INTERNET ADOPTION IN POST-COMMUNIST COUNTRIES:
A PROPOSED MODEL FOR THE STUDY OF INTERNET DIFFUSION
DAN1ELA V. DIMITROVA
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNrvERsnr of Florida
I greatly appreciate the guidance and support of Dr. Sylvia Chan-Olmsted, my
dissertation chair. Thanks are also due to all of my other committee members. Professor
Wayne Wanta, now at the University of Missouri, was instrumental to the beginning of
this project, and his guidance with the requisite job search at the end was invaluable. Dr.
Kurt Kent was always available for help and advice, both related and unrelated to my
dissertation research. Professor Mindy McAdams believed in my success and contributed
to my development as a teacher and scholar. Last but not least, I want to thank my
external member, Dr. Rich Beilock, who led me to my dissertation idea and also kept me
on my toes, helped with the data collection, offered valuable methodological assistance,
all with a sense of humor.
I thank my mentor, Dr. Lynda Lee Kaid, who was always there for me and whose
passion for research served as a wonderful example.
I would like to thank my parents, Velitcka Ivanova Boytcheva and Vesselin
Dimitrov Boytchev, for their endless support and encouragement in all my endeavors.
Their love for learning and belief in my success is my deepest inspiration.
Thanks go to all my dear friends here in Gainesville and around the world.
Finally, I wish to thank Alexander, who firmly stood by me all the way through this
challenging process. His love and support are invaluable.
TABLE OF CONTENTS
LIST OF TABLES vi
LIST OF FIGURES vii
1 INTRODUCTION 1
Internet Significance 1
Economic Contributions 2
Political Contributions 3
Technological Contributions 4
Social Contributions 5
Other Contributions 6
Post-communist Countries 7
Transition Progress 7
Geographic Regions 9
Economic Inequalities 9
Need for Research of Internet Diffusion 10
Research Method 16
Dissertation Outline 17
2 INTERNET AND SOCIETY 18
Development of the Internet 18
The Invention of the Internet 18
Internet Growth and Global Expansion 21
Internet and Political Development 25
Democracy and the Internet 26
Free Press and the Internet 29
Internet and Economic Development 31
Lower Production and Distribution Costs 32
The Internet Economy 33
Global Markets 34
3 LITERATURE REVIEW 37
Diffusion of Innovations 37
Basic Generalizations 38
Technology Innovation Attributes 39
Cluster Innovations 42
Types of Adopters 43
Other Communication Technologies 46
Levels of Internet Adoption 48
Other Considerations for Internet Adoption 49
New Media Technologies Research 50
Economic Factors 51
Political Climate and Policy 57
Audience Characteristics 69
Cultural Factors 72
Conceptual Framework 76
Further Thought 77
4 METHODS 82
Research Design 82
Data Collection 84
Operational Definitions 85
Economic variable 85
Political climate and policy variables 86
Technology/Infrastructure variable 88
Audience variables 89
Cultural variable 90
Dependent variable 90
The Model 95
Statistical Procedures 96
Multiple Regression Technique 96
Stepwise Regression 99
Proposition 1 100
Proposition 2 100
Proposition 3 101
Proposition 4 101
Proposition 5 102
Proposition 6 102
Methodological Notes 102
5 RESULTS 107
Descriptive Analysis 107
Internet Users 107
Gross National Product 109
Telecommunications Privatization 1 1 1
Bivariate Correlations 112
Regression Results 113
Statistical Assumptions 113
Hypotheses Testing 115
Final Model 119
Tobit Estimates 124
6 DISCUSSION 126
Discussion of Descriptive Analysis 128
Regional Variations 128
Growth of Internet Use 130
Discussion of Hypotheses 2 through 6 134
National Income 134
Telecommunications Privatization 138
Refined Conceptual Framework 147
7 CONCLUSION 151
Theoretical Implications 153
Applied Implications 155
Internal validity 160
External validity 163
Suggestions for Future Research 165
LIST OF REFERENCES 169
BIOGRAPHICAL SKETCH 184
LIST OF TABLES
1-1. Internet Hosts per 10,000 population in 1995 and 1999 15
4-1. Definition of variables in the proposed model of Internet diffusion 83
4-2. Correlation matrix for the continuous independent variables 105
5-1. Internet users per 10,000 people in 1999 109
5-2. Descriptive statistics of variables 1 10
5-3. Pearson correlations between dependent and independent variables 113
5-4. Regression results for Internet users 1 16
5-5. Summary of hypothesis testing 118
5-6. ANOVA Table for Complete Model 120
5-7. ANOVA Table for Model 2 121
5-8. ANOVA Table for Model 3 122
5-9. ANOVA Table for Model 4 123
5-10. Tobit estimates for final model 125
LIST OF FIGURES
1-1. Internet hosts across world regions 12
2-1. Internet timeline 21
2-2. Total number of Internet hosts 22
3-1. Internet hosts across income regions 53
4-1. Graphic model 95
5-1. Distribution of Internet users across countries 1 14
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
INTERNET ADOPTION IN POST-COMMUNIST COUNTRIES:
A PROPOSED MODEL FOR THE STUDY OF INTERNET DIFFUSION
DANIELA V. DIMITROVA
Chair: Dr. Sylvia Chan-Olmsted
Major Department: Mass Communication
This dissertation proposed and tested a five-dimensional theoretical framework to
explain the variations in Internet use across the post-communist countries. The
framework included economic, political climate and policy, technology/infrastructure,
cultural, and audience factors. Three factors emerged as critically important: economic,
political, and infrastructure factors. Cultural factors seemed to have partial impact. These
findings suggest that the traditional country-level indicators of economic wealth and
technological infrastructure remain important determinants of Internet use in the
countries of Eastern Europe and the former Soviet Union. The most significant
determinant, however, was level of democratization.
The results of the multiple regression analysis reported in this study indicate that
democratization, teledensity, and GNP per capita were the three most important factors
positively related to Internet use in the post-communist countries. Being predominantly
Muslim had a negative effect on Internet use while being Western Christian (Protestant or
Catholic) seemed unrelated to Internet adoption. Neither length of telecommunications
privatization nor education level appeared significant in this analysis.
Thus, the results of this study shed light on the macro-level indicators that affect
Internet adoption in the post-communist countries. These have important implications for
policy makers at the local, national, and international level. The proposed Internet
diffusion model may be applicable to other regions, but this analysis focused only on the
28 post-communist countries.
Human history has witnessed the rise and fall of many new technologies. The
Internet is often described as the most revolutionary new technology ever and its growth
in every country around the globe is remarkable as well as irreversible. The Internet is not
just a technological innovation; it is a unique, fascinating, multifaceted network that
transcends national boundaries. It affects people, communities, organizations, and
countries around the world. In this dissertation, Internet adoption is examined in a unique
region of the world-the post-communist countries. 1 The dissertation proposes a five-
dimensional analytical framework to account for differences in Internet diffusion in those
countries. Testing the five-dimensional framework identifies the main determinants of
Internet adoption and illustrates how these determinants affect Internet use in the post-
The Internet constitutes "at once a world-wide broadcasting capability, a
mechanism for information dissemination, and a medium for collaboration and
interaction between individuals and their computers without regard for geographical
location" (Leiner et al., 1997). The Internet has affected society in a variety of ways as a
1 The post-communist countries are defined as the countries of Eastern Europe, the
former Soviet Union, and Mongolia. Specifically, the study examines Albania, Armenia,
Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic,
Estonia, Georgia, Hungary, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Macedonia
(FYROM), Moldova, Mongolia, Poland, Romania, Russia, Slovak Republic, Slovenia,
Tajikistan, Turkmenistan, Ukraine, Uzbekistan, and Yugoslavia (Serbia and
result of its multidimensional functions (Castells, 1996; ITU, 1999; Newhagen & Rafaeli,
1996; OECD, 1998b). One of the most fascinating aspects of Internet diffusion is its
impact on a global scale. The Internet has exercised a tremendous impact on the
economic, political, technological, and social development of countries (IMF, 2000a;
Mitchell, 1995; World Bank, 2001). The positive contributions of Internet adoption in
each of these four areas are reviewed below.
One of the more compelling arguments made to encourage global Internet
diffusion is that countries (developing countries in particular) can improve their
economic status with the adoption of this technology (Hanson & Narula, 1990). Clearly,
the Internet affects the economic situation in a country as it facilitates international trade,
lowers production and distribution costs, optimizes productivity within and between
companies, and offers leapfrogging possibilities for less developed countries.
The Internet facilitates international trade and thus allows nations to increase
exports and imports of goods to and from other countries (The new economy, 2000). It
allows better integration of national markets both internally and
externally/internationally, and also increases possibilities for economic decentralization
(Maddock, 1997). With the improvement of Internet security, e-commerce is expected to
increase and more national exports and imports will be likely (DePrince & Ford, 1999).
In addition, the Internet lowers production and distribution costs (DePrince &
Ford, 1999; Guthrie & Austin, 1996; The new economy, 2000). These lower costs can
have a positive effect on the internal economic situation. The Internet intensifies price
competition among producers, which leads to lower prices for consumers (Guthrie &
Austin, 1996). Labor productivity is increased, and searching, distribution, and
transaction costs tend to drop (DePrince & Ford, 1999).
It has been argued that the Internet optimizes productivity within and between
companies (Maddock, 1997; Malecki, 1997, 2000). The development of a better
telecommunications infrastructure in general, Maddock (1997) argues, increases
productivity and competitiveness of local companies. In addition, telecommunications
development facilitates economic growth by increasing market efficiency (Maddock,
1997). Finally, better telecommunications can improve management within corporations
(Daly & Miller, 1998; Maddock, 1997).
Another economic impact of higher Internet penetration is the leapfrogging effect.
Leapfrogging is the ability of countries that are technologically behind suddenly to skip
generations of intermediate technology and adopt the latest one. The adoption of the
latest technology is seen as beneficial to countries as they can succeed in catching up
economically with more technologically advanced societies (Singh, 1999).
In addition to the economic benefits, the Internet can serve as a tool for enhancing
democratic governance worldwide. A number of scholars have discussed how the Internet
can affect the political situation within a country (Ahmann, 1998; Godwin, 1998; Poster,
1995). From the early days of the ARPANET (Advanced Research Projects Agency
Network), people envisioned the Internet's expansion to a worldwide, borderless network
(Rogerson & Thomas, 1998). This global network can strengthen democracy in at least
two ways: first, people can stay better informed and thus can make better choices;
second, the Internet offers citizens a global forum for free expression and exchange of
ideas (Perrit, 1999; Poster, 1995, 2001).
The Internet provides citizens in any country with the opportunity to stay better
informed and thus learn more about options for political action and democratic
governance. The information available online is rich and is difficult to censor. The
sources of information are also numerous, ranging from established media corporations to
independent journalists to regular citizens publishing online. People can find an
enormous amount of information on any topic that interests them. Better informed
citizens, arguably, can make better decisions in society and more informed political
Another way in which the Internet can strengthen democracy on a global level is
by offering citizens a forum for free expression and exchange of ideas with like-minded
people. In most countries, people are free to go online and express their views to a global
audience. They can also search for and communicate with like-minded people around the
globe. Such online communication and uncensored expression reaches a large number of
people and extends individual freedom. It also allows citizens to organize collective
action and thus influence public policy.
It is important to note that control of the Internet by one government or corporation
is unlikely. Even though regulation can limit the use of the Internet to some degree, it is
quite difficult to enforce such regulation on a global scale.
Furthermore, the Internet brings technological growth to a country as it can
strengthen its overall telecommunications development. Maddock (1997) argues that
telecommunication causes development progress in several ways. First, it leads to the
creation of at least one leading sector of the economy in the country. Second, it
accelerates diffusion of other technologies and thus allows faster catch-up for less
developed countries. In today's day and age, it is hardly questionable that the Internet is
critical for the technological advancement of a nation and has become an indispensable
part of the modern telecommunications infrastructure.
The Internet also brings about social change. With the advent of the Internet, we
may be coming closer to what Marshall McLuhan conceived as "the global village"
(McLuhan & Powers, 1989). This is another reason why it is important to research
Internet diffusion. The emergence of a global community is facilitated by the Internet.
People from all nationalities and various backgrounds can form communities online.
Boundaries and distances between countries shrink in the virtual world. Thus, the Internet
(through email, chats, and bulletin boards) brings people closer, regardless of geographic
location. These online applications redefine social relationships within countries. The
effect of virtual communities can increase since email remains one of the most popular
Winner (1997) discusses the idea that technological innovations lead to social,
cultural, and political transformation. He argues that "technical innovations of any
substantial extent involve a reweaving of the fabric of society, a reshaping of some of the
roles, rules, and relationships that comprise our ways of living together." New
technologies, then, should be studied closely and not only at one point in time. Their
effect on the relationships in society should also be examined and their long-term impact
followed, if we are to understand adequately the whole diffusion process and its
In addition to the benefits noted above, the global expansion of the Internet is
beneficial to countries in the areas of education and health (ITU, 1999; World Bank,
2001). The Internet as an unlimited and readily accessible information resource can be
useful for data retrieval on a global level (Sadowsky, 1993). It can also facilitate better
service delivery and learning through distance education (World Bank, 2001).
Finally, intellectual curiosity makes it a worthwhile effort to study the expansion of
the global network of networks. Press has put it well (1997):
Tracking the diffusion of the Internet is a daunting task because it is growing
rapidly, is global, and expands organically, at the edges and internally, without
central control. Still, business people, policy makers, and capacity planners are
better off with approximate data than none at all. . . . Over the years, busy humanity
has covered the globe with cities linked by railroads, highways, telephone lines,
power grids, canals, and so forth, and we are now weaving digital communication
links-the nervous system. I suspect that curiosity and aesthetics motivate the
people tracking the global diffusion of the Internet as much as profit.
In this dissertation, the challenging task of tracking and explaining Internet
adoption in a unique region of the world is pursued. The study examines Internet
adoption in the countries of Eastern Europe and the former Soviet Union and analyzes the
multivariate relationships to explain their levels of adoption. The dissertation proposes a
five-dimensional conceptual framework to account for the differences in Internet
diffusion in the post-communist countries. The study draws on research from the United
States, Western Europe, and Latin America as well as cross-country Internet diffusion
literature. It proposes and tests a comprehensive five-dimensional theoretical model.
Based on the results, a refined conceptual framework is retrieved. Discussion shows how
that refined model can be employed in predicting and modifying future Internet adoption
at the country level. The rest of this chapter briefly describes the post-communist
countries and then outlines the significance of region-specific research on Internet
The post-communist countries examined in this research include the former Soviet
Union republics, Eastern Europe, and Mongolia. Specifically, they include the following
28 nations: Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria,
Croatia, Czech Republic, Estonia, Georgia, Hungary, Kazakhstan, Kyrgyzstan, Latvia,
Lithuania, Macedonia (FYROM), Moldova, Mongolia, Poland, Romania, Russia, Slovak
Republic, Slovenia, Tajikistan, Turkmenistan, Ukraine, Uzbekistan, and Yugoslavia
(Serbia and Montenegro). These countries represent a unique region of the world. Since
the end of the Cold War they have been undergoing a transition from totalitarian regimes
to democratic societies. A Freedom House report titled Nations in Transit examines the
transition in the region, which is characterized by trends toward building civil society and
market economy (Karatnycky et al., 1997). The report shows that the post-communist
countries can be divided into three distinct groups, based on progress in these two areas
(Karatnycky et al., 1997). The members of the three groups are listed below.
The first group includes the leaders in the transition process: Poland, the Czech
Republic, Hungary, Slovenia, Estonia, Latvia and Lithuania. These countries are
classified by the Freedom House based on their 1996 survey as "liberal democracies, or
polities that are well on their way to democracy, with vibrant civil societies, well-
established rule of law, and market economies" (Karatnycky et al., 1997, 17). These six
nations, no doubt, have made considerable progress and are leaders in the transition
process in the region.
The next group of post-communist countries was classified as the intermediary
group. It consists of countries that have made some progress towards the goal of building
strong democratic societies, but the transition has not been as quick or smooth as in the
leading countries noted above. Albania, Armenia, Bulgaria, Croatia, Georgia,
Kyrgyzstan, Macedonia (FYROM), Moldova, Romania, Russia, the Slovak Republic, and
Ukraine are members of the intermediary group. Bosnia and Herzegovina and Yugoslavia
(Serbia and Montenegro) are also members of this group. There is considerable variation
across these countries, both in terms of political development and economic progress.
The third group among the post-communist countries consists of those that have
been slow to change, relative to the rest of the transition societies- Azerbaijan, Belarus,
Kazakhstan, Tajikistan, Turkmenistan, and Uzbekistan. The Freedom House report posits
that the first and the third group are more stable while the middle group is less stable as
countries belonging to that group are more fluid and can transition to the next levels
rather quickly (Karatnycky et al., 1997). However, the three groups are distinctly
different from each other and are likely to remain different over time. In other words, the
distinctive features of the three groups of transitional societies are stable. An interesting
observation is that the countries in the three groups are clustered together geographically.
Also, "the intermediate countries deserve particular attention, as their variable legacies
and half-hearted reforms do not imply any clear-cut outcomes" (Karatnycky et al., 1997,
Another grouping of the countries in the region is given by the United Nations
Development Programme (UNDP, 1999), which uses six regional groupings in its Human
Development Report for Central and Eastern Europe and the Commonwealth of
Independent States (CIS), based primarily on geographic location.
The post-communist countries can be divided into 6 regions based on their
geographic location: Central Europe, Eastern Europe, Caucasus, Baltic states, Western
Former Soviet Union, and Central Asia.
Probably the most advanced group of countries is located in Central Europe. This
group includes the following states: the Czech Republic, Hungary, Poland, the Slovak
Republic, and Slovenia. The second geographic group is Eastern Europe. The Eastern
European countries are Bulgaria, Croatia, Macedonia (FYROM), Slovenia, Romania, and
Yugoslavia (Serbia and Montenegro). The Caucasus region includes Armenia,
Azerbaijan, and Georgia. All of these counties are former Soviet republics. The Baltic
states have also made considerable strides in the post-Cold War transition. These states
include Estonia, Latvia, and Lithuania. The Western FSU (Former Soviet Union)
incorporates the following states: Belarus, Moldova, Russia, and Ukraine. Finally,
Central Asia includes the republics of Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan
Using national income as a criterion, the post-communist countries can be divided
into four groups, according to the 2000 World Development Report of the World Bank,
which is based on 1998 data. The four groups are low income (less than $760 GNP per
capita), lower middle income ($761 -$3,030), upper middle income ($3,031 -$9,360), and
high income ($9,361 or more).
The low income countries in the region are Albania, Armenia, Azerbaijan, Bosnia
and Herzegovina, Kyrgyzstan, Moldova, Mongolia, Tajikistan and Turkmenistan. The
next category-lower middle income-includes the majority of the countries: Belarus,
Bulgaria, Georgia, Kazakhstan, Latvia, Lithuania, Macedonia (FYROM), Romania,
Russia, Ukraine, Uzbekistan, and Yugoslavia (Serbia and Montenegro). Upper middle
income states are Croatia, the Czech Republic, Estonia, Hungary, Poland, and the Slovak
Republic. The only high income country in the region is Slovenia (World Bank, 2000).
In conclusion, the post-communist countries have made different progress in the
post-Cold War transitional period (de Melo & Gelb, 1996). Variations both in political
and economic development exist (EBRD, 1997).
Need for Research of Internet Diffusion
Understanding the process of Internet adoption at the county level is, first and
foremost, critical for formulating public policy. National policies can contribute to
accelerated Internet adoption, which can be beneficial to the country. As noted earlier, the
Internet can contribute to national development in several ways. Internet adoption has a
positive economic impact overall. The utility of the Internet as a political tool, which
allows people to stay better informed and participate more fully in the political processes,
has also been discussed. Country-specific research on Internet diffusion can also be used
to enhance the technological development of the country, as outlined above. Finally, the
Internet has important social functions, as it allows individuals to create online
communities and interact with people around the globe, regardless of geographic
boundaries. The Internet deeply impacts societies. Therefore, it is critical to better
understand the process of its adoption. This will enable policy makers to exploit the
Internet's full capabilities (Maherzi, 1997; Sadowsky, 1993; World Bank, 2001).
The dissertation focuses on the aggregate level of Internet adoption. Various
studies in multiple disciplines have examined the topic of Internet diffusion at the
individual level. Most research has been conducted in the industrialized nations. Research
on Internet adoption and diffusion in the United States in particular has been quite
extensive (Atkin et al., 1998; GVU, 1998; Lin, 1999; Lindstrom, 1997; NTIA, 1995.
1998, 1999; Pew, 1995; USIC, 2000) and the number of studies looking at Internet
adoption and usage keeps growing.
The focus of Internet studies in general has been primarily on the United States
(Daly & Miller, 1998). Literature exists also on Internet adoption by other industrialized
nations, such as The Group of Eight (G-8) nations (Canada, France, Germany, Italy,
Japan, Russia, United Kingdom, and United States) and OECD member countries
(Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Finland, France,
Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Korea, Luxembourg, Mexico,
Netherlands, New Zealand, Norway, Poland, Portugal, Spain, Sweden, Switzerland.
Turkey, United Kingdom, and United States) (Hargittai, 1999; McElhinney, 2001). Yet
few studies have examined the process of Internet adoption in other countries.
As the birthplace of the global information superhighway, the United States
remains one of the countries with the highest Internet penetration (ITU, 1999; Pitkow,
1996; USCD, 1998; USIC, 2000). Other developed countries (notably Scandinavian
nations) are also leaders in Internet adoption. Gunarante (2001) identified three "global
centers" where information technology is concentrated: NAFTA center, EU center, and
Asian-Pacific center. Countries that do not belong to any of those centers are considered
on the periphery of the Information Society (Gunarante, 2001).
The global distribution of Internet hosts shows that more than 88 percent of the
hosts in 1999 were located in North America and Europe (ITU, 1999). The members of
the European Union clearly have higher Internet penetration than the rest of the world. As
Figure 1 illustrates, South Asia, North Africa, the Middle East, and sub-Saharan Africa
exhibit lower levels of Internet adoption (ITU, 1999; World Bank, 2000). The countries
where extensive Internet research exists tend to have higher Internet penetration. Figure
1-1 shows the unequal distribution of Internet hosts in different regions of the world.
Internet Hosts by Region
South Asia Middle East* North Sub-Saharan Africa East Asia & Pacific Latin America & Europe & Central Asia European Union
Figure 1-1. Internet hosts across world regions. Source: World Bank, 2000.
The obvious discrepancy between the industrialized countries and the rest of the
world makes it not only interesting but also critical to study the determinants of Internet
diffusion. It is very important to examine whether and how the "global information
highway" is adopted by other countries, especially considering the possibility that the
Internet not only bridges, but also widens the gap between rich and poor countries (Tele-
haves, 1996; WIPO, 2001). Furthermore, cross-cultural studies allow for identifying
regional differences in the adoption of information technologies.
Studies of Internet adoption at the country level will not only provide insights about
the process of Internet diffusion across countries but also show whether this process
varies by certain country-level characteristics. Our understanding of Internet diffusion in
non-Western countries is still very limited. As shown above, differences in levels of
Internet penetration between industrialized countries and the rest of the world are quite
significant. These differences cannot be fully explained by the existing literature. This
dissertation extends current literature by offering insights on Internet trends in the post-
Better understanding of how new technologies are adopted by post-communist
countries can help development in those countries. Even though the direct causal
relationship between Internet usage and development/economic growth has been debated,
studies have shown consistently a strong positive correlation between telecommunication
services and country-level economic indicators such as per capita income (Arnum &
Conti, 1998; Elie, 1998; Hargittai, 1999; Singh, 1999). The process of Internet adoption
remains not fully understood. It has become clear that per capita income plays an
important role, but few studies have examined what other factors affect Internet adoption
at the macro/societal level. This study adds to current literature as it proposes a more
comprehensive model of Internet diffusion.
This dissertation focuses on Internet adoption in the former Soviet bloc. The post-
communist countries are of particular interest for several reasons. First, they present a
unique case study: countries that were subjected to communist rule for 40 to 70 years
give researchers a chance to follow the transition of telecommunications in post-
communist societies (Katchanovski, 2000). As Rose (2002, 33) notes, "Internet access is
especially important in the transition countries, because the transition process is about
opening up a country to the world."
The post-communist countries share at least 40 years of Soviet influence, which
makes them a different cluster for research. They share certain objective characteristics,
such as high literacy rates and high educational levels (UNDP, 1999). At the same time,
they differ in some cultural aspects, such as religious beliefs and historical experience
(Katchanovski, 2000). The focus on one particular region of the world allows researchers
to detect more intricate relationships among variables and gain insights about the
magnitude of importance of country-level characteristics.
The post-communist countries present the opportunity for a case study of the
Internet adoption process among a group of relatively similar countries. It is interesting to
examine whether and how countries that had relatively similar technological, political,
and economic levels have reached significant differences in the area of new information
and communications technologies, such as the Internet. The post-communist countries
can either follow the models of Internet adoption of other countries—the industrialized
countries, for instance—or exhibit a different path of Internet adoption due to their unique
socio-economic development. The results of the study advance knowledge of different
patterns of global Internet diffusion.
Further research on Internet adoption in the region is needed because it has been
identified as the next area where an Internet boom will be seen (Arnum & Conti, 1998;
ITU, 1999; USIC, 2000). This upward trend makes research on Internet adoption in the
post-communist countries even more timely and important.
Table 1-1. Internet Hosts per 10,000 population in 1995 and 1999.
Hosts in 1995
Hosts in 1999
Bosnia and Herzegovina
Yugoslavia (Serbia & Montenegro)
Source: World Bank, 2000.
internet use in the post-communist countries has increased exponentially over the
past decade (ITU, 1999; Magyar & Karvalics, 2001). Table 1-1 illustrates the growth of
Internet usage in the post-communist countries by showing the increase in the number of
Internet hosts. From 1995 to 1999, that number has doubled at the least. In the case of
Belarus, for example, Internet hosts increased by 3,850 percent. Lithuania, for example,
had only 1.23 hosts per 10,000 in 1995, but the number increased 2,376 percent to 30.45
hosts per 10,000 in 1999. Bulgaria had a similar number of hosts in 1995-1.26 per
10,000-but the increase in 1999 was relatively small compared with Lithuania-to 1 1.9
hosts only. These examples illustrate that the number of Internet hosts per capita in post-
communist countries remains uneven and the growth rate across countries varies widely.
Among the Central and Eastern European nations, Slovenia, the Czech Republic,
Slovakia, Hungary, and Poland have higher Internet penetration than the rest of Eastern
Europe. Among all post-communist countries, Slovenia and Estonia seem to exhibit
higher rates of Internet adoption (CDT, 2000; ITU, 1999; World Bank, 2000). These
variations in Internet penetration among countries with relatively similar socio-economic
developments in the post-Cold War years suggest that there are multiple factors that
affect Internet adoption on a country-level basis. The main contribution of this
dissertation is to identify a group of variables that constitute the most important
predictors of Internet diffusion in this particular region of the world.
The dissertation employs aggregate data to determine the multivariate relationships
between Internet adoption and a number of explanatory variables. 2 These variables fall
into five categories: economic, political, technological, cultural, and audience factors.
The study design reflects the theoretical framework discussed in the literature review
chapter. The dissertation employs t-test, multiple regression, and Tobit analysis to
The study is limited by data availability. It relies on the latest data published by various international
organizations. Data paucity accounts, in part, for why Internet adoption in the post-communist countries
has not been well researched so far.
determine the statistical significance of a set of explanatory variables. A more detailed
explanation of the methods is provided in Chapter 4 of the dissertation.
After discussing the need for research in Chapter 1, the dissertation continues to a
more detailed explanation of the significance of the Internet for the political and
economic growth of society in Chapter 2. Chapter 2 also offers a brief historical overview
of the major events that led to today's Internet. Next, Chapter 3 reviews relevant
literature and explains how it is related to the present study. Chapter 3 also describes the
comprehensive five-dimensional theoretical model proposed and tested in the
dissertation. Chapter 4 explains the methods used and describes the study design, data
collection, operational definitions, and scales of measurement. The statistical results are
analyzed next in Chapter 5. The discussion of the results follows in Chapter 6. Chapter 7
summarizes the entire dissertation. It also addresses broader implications and limitations
of the study. Suggestions for future research and theoretical and practical consequences
also are offered.
INTERNET AND SOCIETY
The Internet's contributions to society are manifold. The Internet is especially
important for the political and economic development of nations. The first section of this
chapter presents an overview of Internet development and growth from the 1960s until
today. A timeline including Internet milestones is provided. After outlining briefly the
history of the Internet, the chapter describes the importance of the Internet for society as
it relates to two major issues: the political process and economic development.
Development of the Internet
Even though the concepts underlying today's Internet were created in the late
1960s, the Internet diffusion on a global level occurred only in the 1990s. Several key
developments that led to the current state of the Internet are presented below.
The Invention of the Internet
The Internet was created in the 1960s as a result of close collaboration within the
American research community (Hafner & Lyon, 1996). Tracing the history of today's
Internet, it is important to note that four significant aspects of the network were
considered from the start: technological, management, social, and commercial (Leiner et
The first documented idea of an Internet was discussed in internal memos by J. C.
R. Licklider at the Massachusetts Institute of Technology (MIT). Licklider described his
idea of a "galactic network" in these memos in 1962 (Leiner et al., 1997). Other
developments such as the first work on packet switching theory took place during the
early 1960s as well (Hafner & Lyon, 1996). The Advanced Research Projects Agency
(ARPA), a research and development organization funded by the U.S. Department of
Defense, began operating in 1967.
ARPA was instrumental in the development of the early Internet. ARPA tested data
transfer across telephone circuits using packet-switching and thus became the first agency
to implement a network based on that technology (Leiner et al., 1997). This network was
named Advanced Research Projects Agency Network-or ARPANET-and became the
forerunner of the Internet (Hafner & Lyon, 1996).
Robert Taylor at ARPA found funding to test this so-called "network experiment"
beginning with a few nodes (Hafner & Lyon, 1996). The first two computers linked on
the ARPANET were at the Network Measurement Center at the University of California
at Loss Angeles (UCLA) and the Stanford Research Institute (SRI) at Stanford
University. The exchange of the first host-to-host message took place on that network in
1969. Soon after that, two more nodes were added: the University of California at Santa
Barbara (UCSB) and the University of Utah. Thus, four host computers were connected
by the end of 1969 into the initial ARPANET (Leiner et al., 1997). These four nodes
were the first realization of the idea of the "galactic network." Today it is common to
refer to the 1969 ARPANET as the earliest existing Internet (Hafner & Lyon, 1996).
From the beginning, the Internet was conceived as a general infrastructure that
would connect multiple independent networks and support numerous and new
applications. It is considered an open-architecture network— a network of interconnected,
independent computers. The structure of the Internet was "foretold" by Paul Baran, an
engineer at the RAND corporation (Hafner & Lyon, 1996). He created the idea of a
distributed network, a digital switching technology to connect computers at various
locations. Baran also came up with the idea to break down the message into small pieces
(packets) that will travel independently over a network and then reconnect before arriving
at their final destination. This is still how messages are sent on the Internet.
A major step toward developing today's Internet was the creation of TCP/IP
(Transmission Control Protocol/Internet Protocol). This protocol meets the needs of an
open-architecture network environment. It was compiled in 1972 by Robert Kahn at
DARPA and Vinton Cerf at Stanford University (Leiner et al., 1997). Basically, the
TCP/IP protocol allows information bits to travel to destinations independently. The
Internet Protocol (IP) is responsible for addressing and forwarding of individual packets
while the Transmission Control Protocol (TCP) is responsible for service features such as
flow control and recovery from lost packets.
It is important to note that the researchers at ARPA, its contractors, and several
universities collaborated closely to create, test, and improve the Internet invention
(Hafner & Lyon, 1996; Leiner et al., 1997). In the post-Sputnik era, money for research
from the U.S. government became abundant. As Hafner and Lyon (1996, 23) note,
science was "the New Frontier." The initial collaboration between the academic research
community and ARPA, an agency within the Defense Department, led to the successful
implementation of the ARPANET (Hafner & Lyon, 1996). Therefore, the Internet
represents a good example of "the benefits of sustained investment and commitment to
research" (Leiner et al., 1997). A number of people contributed to the development and
improvement of the initial Internet, but the original research and implementation
happened mostly under ARPA's funding umbrella in the 1960s (Hafner & Lyon, 1996).
Below, some milestones in Internet development are shown.
1962-Primary idea of galactic network
1967-ARPANET development begins
1969-First four host computers connected over the ARPANET
1972-First public demonstration of ARPANET
1972-First email application created at BBN
1973-First international connections to ARPANET created in England and Norway
1974-First use of the term "Internet" in a conference paper by Cerf and Kahn
1978—Trials for a private system, Telset, begin in Finland
1983— Domain Name System developed at the University of Wisconsin
1986-NSFNET replaces the ARPANET
1 989-- World Wide Web invented
1993— Mosaic, the first Web browser for personal computers is created
Figure 2-1. Internet timeline.
Internet Growth and Global Expansion
The previous section shows that initially the Internet was nothing more than a
network to used internally within a particular agency. Access to that network was limited
to a handful of people and information about it was sparse. The first public demonstration
of the network happened only in 1972. Even at that time, however, people envisioned its
expansion to a worldwide, borderless network (Rogerson & Thomas, 1998). This was not
yet technically possible though. The Web browser— the graphical interface of the World
Wide Web— was created in 1993. The first graphical browser Mosaic was the precursor of
Netscape and later Internet Explorer. It encouraged faster diffusion of the Internet around
the world (Winston, 1998). Figure 2-2 shows the growth of Internet hosts over the years
on a global scale.
Internet hosts worldwide
# a j>j> j> * + j- j> s jp * # # j>j> & $ A A * * *j> # # # * * a *
# v* jf c^ 5f v* y c^ / v* y <P 4 ^ y/f//(P / v* *W<f* ftfWl #
Figure 2-2. Total number of Internet hosts. Source: Internet Software Consortium, 2000.
Clearly, the use of the Internet increased dramatically after the free distribution of
the Web browser. The World Wide Web was developed by Tim Berners-Lee at CERN in
Switzerland. The World Wide Web (WWW) uses HTML (hypertext markup language),
which incorporates text and graphical elements as well as hyperlinks. The introduction of
the World Wide Web made it easy for non-technical persons to use the computer
network. As the inventor of the World Wide Web acknowledged, "transferring
information was too much of a hassle for a non-computer expert" before that (Berners-
Lee, 1999, 18). Berners-Lee' s goal was not only to make it easier to use the Internet, but
also to create a system in which different computers with different software could
connect and "talk" to each other and thus enable researchers to share their work quickly
and easily. Berners-Lee's first formal proposal for funding was submitted in March 1989
at CERN, but received no feedback. Another proposal followed with the same result. As
Berners-Lee wrote (1999, 27), "explaining the vision of the Web to people was
Finally, the World Wide Web was released in 1993. Like the Internet on which it
runs, the Web has no central location as the online information is distributed: i.e.,
documents are stored on many computers all over the world. The Internet has no main
node, so it has an infinite storage capacity as a result.
The global growth of the Internet has been impressive and often labeled
"revolutionary" and "phenomenal." In fact, it has been argued that the Internet is the
fastest growing communications technology ever (WIPO, 2001). In 1990, only 22
countries were connected to the Internet, compared to 226 countries in 1999. The number
of countries connected to the global network increased tenfold in less than 10 years. As of
September 2002, there are 605.60 million online users worldwide (Nua, 2002).
Even though the global expansion of the Internet has been accelerating, disparities
between regions and countries do exist. The United States and North Western Europe
have the lion's share of the Internet (ITU, 2000). The Internet is expected to become less
U.S. -centric though. Forecasts show that Internet growth in the United States will level
off by 2002, and most continuing growth will be observed in Western Europe and
developed Asia (Bieler & Stevenson, 1998).
In the Asia-Pacific region, Australia, Japan, and New Zealand are clearly the
leaders in Internet adoption. Hong Kong, Singapore, South Korea, Taiwan, and China
also exemplify fast Internet growth rates (USIC, 2000). The rest of the region and South
Asian countries in particular lag behind in Internet usage. Japan is expected to continue
dominating the Asian Internet market (Bieler & Stevenson, 1998).
In Europe, there are disparities in Internet penetration across East- West and North-
South lines. Internet penetration rates in Sweden, Norway, Denmark, and Finland
surpassed 35 percent in 1999 (USIC, 2000). Italy, France, and Spain had relatively lower
Internet use compared with the Northern European and Scandinavian nations. With the
exception of Slovenia, the countries of Eastern Europe are further behind. Analysts
forecast, however, that Eastern Europe and the former Soviet Union are likely to exhibit
high Internet growth rates (Bieler & Stevenson, 1998; ITU, 1999).
Projections show that together with Eastern Europe, Latin America will be one of
the regions to experience substantial increase in Internet usage (USIC, 2000). Internet
adoption in Latin America and the Caribbean is growing steadily. Brazil, for instance, has
one of the fastest growing ICT markets in the world (World Bank, 2001). Mexico and
Argentina, in addition to Brazil, are leaders in Internet usage in the Latin American
In Africa, Internet growth rates have been relatively low. According to 2000 data,
there were 1.5 million Internet users in Africa (USIC, 2000). About 1 million of those~or
two-thirds of all Internet users, however, were located in South Africa (USIC, 2000). In
the summer of 2000, more than 300 million people worldwide were using the Internet
regularly (USIC, 2000). Yet only 0.6 percent of the people living in developing countries
had access to the Internet (USIC, 2000). As a recent report from the Center for
Technology and Democracy (2000) notes, Internet disparities bring a danger of dividing
the countries in the world into "information rich" and "information poor."
The Internet is constantly evolving and expanding. It is hard to control the Web,
both in terms of access and content, unlike other media (Perrit, 1999). Regulation has
been formulated within different countries, including China, Singapore, and Turkey, to
try to limit freedom of expression and censor the Internet (Cortez, 2000). Legislation to
restrict Internet in some way (by censoring online information, licensing Internet
providers, etc.) has been proposed in Australia, Chile, Great Britain, and South Africa
(Cortez, 2000). In China, for example, cyber cafes must obtain a license from the state
and ISPs are required to "register their customers with the authorities" (Cortez, 2000). In
addition, Web sites containing subversive information are blocked. However, it is
generally hard to implement Internet regulation due to the global and changing nature of
the network (Perrit, 1999).
The World Wide Web and the Internet as a whole offer unique opportunities for
nations, both in terms of political and economic development. These opportunities are
Internet and Political Development
A number of studies have discussed the potential of the Internet to enhance
democratic governance. Ahmann (1998), for instance, looked at how the Internet can
increase political participation in South Africa. She found that political information
online is abundant and that the Internet can easily be used for political education within a
country (Ahmann, 1998).
The Internet, then, can strengthen the democratic process by serving as a vehicle
for political education of citizens. This is achieved not only by keeping the citizens better
informed (with access to various sources and media online), but also by providing a
public forum for communication and exchange of information with other like-minded
people. Thus the Internet can serve as a channel for mediated interpersonal
communication and community formation.
Interestingly, in the early 1960s J.C.R. Licklider, one of the key figures in the
invention of the Internet, conceived it as a network that would, indeed, allow citizens to
participate in the political process more actively. He envisioned people on the global
network attending a "giant teleconference" (Hafner & Lyon, 1996, 34).
Perrit (1999) discusses the role the Internet can play in strengthening both national
and global governance. He sees the potential of the Internet in four specific areas, all of
which lead to strengthening international cooperation. Perrit (1999) argues that the
Internet can strengthen international law and can also empower and improve local non-
governmental organizations (NGOs). The Internet has the potential to support the
international security system. Finally, the Internet can strengthen the economic
interdependence between countries (Perrit, 1999).
Democracy and the Internet
Many scholars at different times in history have tried to define what democracy
means. There is no single clear-cut definition that encompasses all characteristics of
democracy. The meaning of democracy varies with time and place. The 2000 political
crisis in Yugoslavia shows the significance of one aspect of a democratic society— free
and fair elections. When the Serbian people realized that Slobodan Milosevic was trying
to manipulate election results, they took to the streets because the mandate of their vote
was not acknowledged.
In addition to free elections, people in a democracy require freedom of expression.
This is also supported by a Serbian example, which speaks to the fact that the nature of
the Internet makes it hard to control. Slobodan Milosevic tried to censor print and
broadcast media during the 1999 Kosovo crisis. Milosevic's government also tried to
suppress media freedom by shutting down those media that supported government
opposition. After being closed down, a popular opposition radio station-Radio B92-used
the Internet to "broadcast" to the outside world. Thus, the Serbian opposition was able to
distribute news and information that was censored by other media. This clearly speaks to
the democratic potential of the Internet. During the Serbian crisis, this new medium
provided one of the few channels for pro-democratic groups in the country to speak to the
outside world while other domestic media were strictly censored by the Serbian
government. As Perrit (1999) notes, "the decentralized nature of the Internet itself . . .
makes it very difficult for . . . governments to control and censure political thought,
speech, and action."
An attempted legislation in the United States also clearly shows the value of a free
Internet. The proposed Communication Decency Act, which was not accepted by the U.S.
Congress, shows that freedom of expression is highly valued (Cortez, 2000). The Reno
vs. ACLU lawsuit exemplifies the value of freedom of expression embedded in the
American constitution. As Cortez (2000) notes, the following international treaties also
recognize freedom of expression as a basic human right: the Universal Declaration of
Human Rights of 1948, the Pact of Civil and Political Rights of 1966, and the Pact of
Economic, Social, and Cultural Rights of 1966. By extension, all countries that are
signatories of these international treaties have to harmonize their national legislation with
the provisions of the treaties (Cortez, 2000).
The Internet has served as a vehicle for communication for anti-government groups
worldwide. In China, Lin Hai was given a two-year sentence for sending email addresses
to an anti-government publication (Cortez, 2000). The Internet provides an outlet both for
reaching other anti-government activists directly and for publishing relevant information.
In the case of Malaysia, the Internet allowed the political opposition that was living on a
remote island to use email to organize anti-government action.
Diversity of opinions is critical for a well-functioning democracy. The more voices
are expressed in a public forum, the better. The value of having more information
available from more sources than before, expressed in a public forum, can hardly be
disputed— it makes democracy stronger (Held, 1995).
The Internet certainly makes it easier to get more information from more sources
than ever before and provides a unique opportunity to communicate online, despite
geographic distances. Perrit (1999) says: "The ease with which people can participate in
cyberspace activities enabled the Internet to grow exponentially with virtually no
governmental oversight. This growth has created a cyber-culture that celebrates freedom
and distrusts traditional political institutions trying to come to grips with the implications
of this profound electronic revolution in information technology."
Poster (2001) discussed Habermas's idea of the public sphere—a "space" where
citizens deliberate and interact to form public opinion— as it relates to the Internet.
Jakubowicz says that the "public sphere is a forum of public debate where citizens can
debate issues of common concern, voice and act on their views and seek to arrive at a
consensus on matters of general interest" (Jakubowicz, 1998, 12). Gaynor (1996) argues
that it is difficult to see if the Internet enhances or fragments the public sphere in a
democracy. Because it offers citizens a venue for expressing public concerns, it does,
indeed, serve as a part of the public sphere. Thus, the Internet enables individuals to
influence public policy through the pressure of public opinion.
Sen (1999a, 10) argues that a democracy requires "the guaranteeing of free
discussion and uncensored distribution of news and fair comment." The importance of a
free press and informed citizenry is discussed next.
Free Press and the Internet
The Internet affects media organizations around the world. The structure of the
Internet makes national boundaries irrelevant to the distribution of media messages. The
Internet also allows people to speak publicly and publish information, just like the
traditional press in the past. Online media, however, reach wider, global audiences
(Perrit, 1999). Thus, citizens can access various sources and free media over the Internet.
Today we seem to take for granted that a free press is a vital part of any democratic
society (Jakubowicz, 1998). But what specific functions do the media perform in a true
democracy? According to McChesney (1999), the press has a responsibility to perform a
public service. A free press, he adds, should encourage diversity of opinions. A
democratic society presupposes pluralism, i.e. the expression of different opinions. Under
Communism, people living in the so-called Soviet-bloc countries not only couldn't
express their opinions freely, but they even felt the pressure to say "the right thing," even
if it wasn't what they believed (Yakovlev, 1989). The state prevented the existence of a
free press by enforcing strict censorship and also having financial control of all media.
The Communist Party also jammed foreign media that broadcast within the Soviet bloc to
prevent diverse viewpoints from reaching the average citizen. The official press was
simply a mouthpiece of the communist government.
In a democratic society, however, the press should not be a mouthpiece of the
ruling party. On the contrary— it should hold the government accountable for its actions.
The press is said to have a watchdog function— to hold the government accountable for its
actions. The media in a democracy should serve the public by ensuring that the
government is not abusing its powers. A free press in a democratic society is a place
where deliberation should happen, according to Nader (Nader, 1998). The press, he
argues, should enliven public debate and engage people into it. Such lively press
functions as part of the so-called public sphere. The public sphere, as mentioned above,
enables individuals to give their input through public debate, which is necessary for a
healthy democracy. If the Internet acts as a space for public debate, then it can empower
Another important basic function of a free press is to keep citizens informed. It has
been argued that citizens on a global level can stay better informed through the Internet.
Thus, citizens can participate better in the democratic process in any country (Sen,
1999b). Sen says that "guaranteeing of open discussion, debate, criticism, and dissent, are
central to the process of generating informed and considered choices" (Sen, 1999a, 9).
This is even more important to people in the post-communist countries who, as
mentioned above, were not allowed to openly state their opinions and engage in free
discussion prior to 1989.
Traditionally, the media have served as gatekeepers in society. They filtered the
information and chose which facts were newsworthy. The media thus used to serve an
important agenda-setting function: they told the public what issues were important. It has
been argued that the gatekeeper role and the agenda-setting role of the traditional press
can be manipulated by governments. The Internet may remove the possibility for such
direct controls. Internet news are not easy to control, as the gatekeeping and the agenda-
setting role can be assumed by the average person.
The Internet transcends national boundaries and withstands governmental controls.
The Internet also allows people to speak publicly and publish information, just like the
traditional press in the past by reaching wider, global audiences (Perrit, 1999). Thus, both
functions are enhanced: (1) citizens can stay better informed (with access to various
sources and free media over the Internet) and (2) citizens can use the Internet as an
avenue for communication and debate in a public forum. This suggests the potential of
the Internet to affect the political process both nationally and internationally, and to shift
the practice of democracy to more active citizenry.
In addition to political development, the Internet also affects economic
development worldwide. The rest of the chapter outlines the most important economic
impacts at the national level.
Internet and Economic Development
There are at least four aspects of the Internet that directly affect developing
countries. First, the Internet has enabled cheaper production and distribution of goods.
Second, the emerging Internet economy has played an important role in reforming
traditional economic structures within countries. Third, the Internet has expanded and
strengthened global markets by enabling more exports and imports between countries.
Finally, it has been argued that the Internet has "leapfrogging" potential for less
Lower Production and Distribution Costs
The 20 th century has witnessed a tremendous growth of technology. One of the
important effects that technology innovations have brought about is the reduced cost of
production (Christensen, 1997; Sadowsky, 1993; The new economy, 2000). Not
everybody agrees that reduced production costs or optimized efficiency is generally good
for society, but still the cost of information exchange today is much lower compared to
the time when computers first started (The new economy, 2000). This is true for other
services as well.
Sadowsky (1993) discusses several ways in which communication technology
affects economic development in society. He contends that the transportation industry as
well as the finance sector would be very different without the use of high-speed
communication and computing technology. Decrease in the price of microelectronics,
satellite technology, fiber optics, and packet-switching, Sadowsky (1993) argues, have in
turn accelerated the adoption of modern computing.
Information technology (IT) has been a major factor in the increased production of
goods and services by U.S. workers (Barua et al., 1999). IT has also made these workers
more efficient. Of the 2.6 percent increase in U.S. labor productivity between 1996 and
1999, more than half was directly related to the information technology sector (USIC,
The Internet intensifies price competition among producers and their suppliers, and
thus leads to lower prices for consumers as well (DePrince & Ford, 1999; Guthrie &
Austin, 1996). Guthrie and Austin (1996) contend that product quality is improved as a
result. DePrince & Ford (1999) argue that labor productivity is increased by Internet
marketing while searching and transaction costs tend to drop. They predict that "as the I-
ECON's share of the total economy rises, the magnitude of the Internet's macroeconomic
impact will also rise" (DePrince & Ford, 1999).
The Internet Economy
The economic impact of the emerging Internet economy is also impressive.
DePrince & Ford (1999) note that the Internet economy has been affecting the following
areas: hardware, software, intermediaries (such as travel and auction companies), and e-
commerce. All these sectors have been transformed, directly or indirectly, by the advent
of the Internet (McKnight & Bailey, 1997).
DePrince & Ford (1999) note that there is a shift from traditional distribution
methods to Internet distribution of specific products and services. They distinguish
between two types of Internet distribution: "Amazonic" and "Dellphic." The former
refers to ordering products online from a company, such as Amazon.com, that maintains
a warehouse of its products. Such transactions can be either public or private, individual
or wholesale. The second distribution method, named after Dell computers, is the
ordering of a product before it has been manufactured. Users make an order; then the
producer creates the product or service and ships it to the consumer/buyer. In this case,
the customer usually responds to promotional efforts by the producer. Both of these
distribution channels have become popular and are affecting the overall economic
structure within and between countries.
The Internet has been seen as a driver of economic boom. The Internet economy
workforce in the United States grew 36 percent from 1998 to 1999 (Barua et al., 1999).
According to the U.S. Commerce Department, out of a 5 percent improvement in U.S.
production of goods and services in 1999, 1.6 percent has been attributed to information
technology (IT) use (Barua et al., 1999). Labor productivity in the United States
increased by roughly 3 percent a year from 1995 to 2001; faster growth was observed in
IT-using industries (Baily, 2001). In other words, studies have shown that IT improves or
at least contributes to economic productivity.
The Internet facilitates electronic transactions. Even though electronic commerce is
not a new phenomenon in itself, the speed of growth has been accelerated by the Internet
(Atkinson & Court, 1998; Dryden, 1998; WIPO, 2001). Online commerce (e-commerce)
is expected to comprise 4.4 percent of the U.S. Gross Domestic Product (GDP) by 2002
(WIPO, 2001). Its growth globally has also been impressive (World Bank, 2001; WIPO,
As Baily (2001) notes, "new technologies are altering the way traditional industries
operate." The Internet has also strongly affected the following areas: financial services,
travel agents, stock trading, computer sales, music CDs, software, and book sales. Some
other types of businesses are likely to be affected by the growing Internet economy as
well. E-bay is one example where the role of the intermediary is eliminated. Other
intermediaries such as stock brokers may disappear as well.
Of course, as DePrince and Ford (1999) point out, not all products and services will
be affected by the Internet economy. They also note that the Internet has a tremendous
impact not only on the micro, but also on the macroeconomic level, as it promotes
improved macroeconomic performance (DePrince & Ford, 1999).
The Dellphic and Amazonic distribution channels described above also work in a
global setting. Business-to-business transactions and international trade are facilitated by
the Internet. DePrince and Ford (1999) conclude that "the emergence of the Internet
economy . . . may well rival the introduction of printing, steam power, the telephone, and
the assembly line as a growth-enhancing innovation."
Electronic communications facilitate international commerce (Bauer, 1994). The
Internet introduces new mechanisms for imports, exports, and trade in general (World
Bank, 2000, 2001). As a global channel, it makes the actual transactions easier and faster.
One type of product that illustrates well the ease of online transactions is information. It
is very easy to buy, sell, or transfer data over the Internet. Information transactions
exhibit economies of scale. The more copies of a data set, for instance, are sold, the
higher the profit margins. Producing an additional copy of the same data set is virtually at
The argument exists that developing countries can "leapfrog" as they adopt new
technologies and thus surpass developed countries. Basically, leapfrogging is the ability
of countries that are technologically behind suddenly to skip generations of intermediate
technology and adopt the latest one. This is seen as economically beneficial to countries
(Singh, 1999). For example, a nation with very low telephone penetration may adopt the
latest technology for mobile phones and thus jump ahead of other states. Such technology
is likely to be cheaper, more efficient, and easier to build. Government policy in
developing countries has been identified as the main barrier to leapfrogging (The new
The term leapfrogging has been used in the literature in two other ways, in addition
to skipping an intermediate technological stage. Singh (1999) argues that the term
"leapfrogging development" has been used to imply that developing countries can skip
stages of development as a results of telecommunications and thus become members of
the postindustrial society. The third way in which leapfrogging has been used is to mean
that telecommunications can itself lead to accelerated development in such countries
(Singh, 1999). Even though this doesn't happen automatically, the possibility exists at
least that adopting latest information and communications technologies (ICTs) and
sophisticated Internet infrastructure will lead to positive economic impact within
Finally, the tremendous impact of the Internet needs to be put into perspective. All
the possible positive changes will not happen automatically (IMF, 2000b; World Bank,
2001). Sale (1999) discusses the negative effects of large displacement of human labor by
the introduction of more sophisticated industrial technologies, such as the Internet.
Solomon (1998) warns us about the possibility of the Internet becoming an electronic
mall. Winner (1997) raises questions about the concentration of wealth and power
"around" new technologies such as the Internet. Another potential drawback of Internet
use is that the Internet allows dominant ideology transfer and consumer life-styles to
other countries. The Internet can also widen the gap between developing and developed
countries. That is one reason for concerns about the uneven Internet diffusion on a global
level. Despite these potential setbacks, it was argued above that the Internet opens doors
to improving democratic governance and economic development of nations worldwide.
This chapter first discusses the diffusion of innovations theory. It provides an
overview of its generalizations and establishes how these generalizations apply to the
Internet. Several different issues that help in understanding Internet adoption from a
diffusion of innovations perspective are identified. Next, the chapter reviews the body of
literature on new media technologies adoption and use. The chapter concludes by
identifying what major factors affect Internet adoption at the country level. A
comprehensive five-dimensional analytical framework is proposed. Finally, potential
interactions between diffusion of innovations and new media technologies research are
Diffusion of Innovations
One way to study the Internet is by using traditional diffusion studies. This is not
an easy task, however. Classification of the Internet as an innovation is difficult because
of its very complex nature. In addition, the diffusion process at the societal level is not
fully understood. Rogers (1995, 5) defined the diffusion of an innovation as "the process
by which an innovation is communicated through certain channels over time among the
members of a social system." An innovation can be an idea, product, value or a new
technology. Diffusion of innovations typology is applied below to one of the fastest
growing new technologies~the Internet.
Diffusion of innovations is one of the most popular social science theories and it
has been used in variety of academic disciplines (e.g., anthropology, sociology,
communications, marketing, geography, education and health). It started in the early
1900s with the writings of Gabriel Tarde, a French philosopher, who described many of
the components of adoption and diffusion in his work The Laws of Imitation (Tarde,
1903). The first systematic diffusion study, however, was conducted at the University of
Iowa in 1943. In their seminal work, Ryan and Gross (1943) examined the adoption of
hybrid seed corn by Iowa farmers. Diffusion research has grown considerably since then.
Diffusion of innovations offers a linear model for the diffusion process, mostly
focusing on the individual level of adoption (Rogers, 1995). It posits that innovations are
typically diffused after going through the following stages: (1) knowledge, (2)
persuasion, (3) decision, (4) implementation (the actual adoption of the innovation), and
(5) confirmation stage. The first two stages in this process resemble a communication
model in which a sender has to get a message across to a receiver. The mass media are
very important at the knowledge stage (but not as important at the persuasion stage) to
get the information out that a new product or technology exists to fulfill a particular need
(Rogers, 1995). '
Diffusion of innovations posits that innovations in general follow an S-curve of
adoption over time. In other words, the first stages of adoption are slower, then we have a
faster increase in the number of adopters (resulting in a steeper slope of the adoption
1 At the knowledge level of adoption, we can distinguish three different levels of knowledge: awareness
knowledge, how-to knowledge, and principles knowledge.
curve), and finally the curve levels off at the later stages. Empirical data for various
technology adoptions support the S-curve typology (Severin & Tankard, 1997).
Technology Innovation Attributes
Everett Rogers, one of the most influential diffusion scholars (Severin & Tankard,
1997), identified five attributes that play a role in whether a particular innovation is
adopted or rejected (Rogers, 1995). They are: (1) relative advantage; (2) compatibility;
(3) complexity; (4) trialability; and (5) observability of the new idea, product or
technology. Rogers later added reinvention as the sixth additional attribute. Reinvention
refers to cases where an existing technology is used for a different purpose than originally
conceived. For example, if we first thought that the computer was supposed to be used
for word-processing, we reinvented its use when we started sending e-cards. Rogers
(1995) contends that 48 to 87 percent of the variability of adoption is explained by the
perceived attributes of an innovation.
The first attribute—relative advantage-refers to the question whether the
innovation is better than what one used before. The perceived relative advantage of the
Internet, at least at first, will be largely dependent on the marketing and promotion efforts
put into mass media campaigns. Compatibility is whether the innovation is compatible
with previous technologies. The next attribute, complexity, refers to the question how
difficult it is to use the innovation. If it is more complex, people will be less likely to use
the innovation. Trialability is whether a person can try out the innovation. Finally,
observability is whether the results of the innovation are readily observable. Clearly there
are extraneous factors "outside" of the individual that will affect these attributes.
Interactive innovations are innovations that depend on the number of people who
have already adopted a particular innovation (Mahler & Rogers, 1999). In other words,
the value of the innovation-its relative advantage in the mind of the potential adopter-
increases if there is a large number of people who are already using that innovation. In
the case of the Internet, people may be more likely to adopt it if they have the ability to
be connected to more people who are using the Internet already. A good example to
clarify this point is email-if your friends have email already, then you can communicate
with them over the Internet. The perceived value of the technology increases as a result.
Several technological innovations have been labeled interactive in nature.
Previous research, however, has not clearly distinguished between interactive and non-
interactive innovations (Mahler & Rogers, 1999). The classic example of the fax machine
shows that if only very few people are using the innovation, its value is relatively low.
Once there is a large number of others that use the fax machine, its diffusion increases
Mahler and Rogers (1999) examine the diffusion of interactive communication
innovations (several telecommunications services) and their adoption by German banks.
They argue that the rate of adoption of such interactive innovations follows a modified S-
curve (Mahler & Rogers, 1999). The adoption process of interactive innovations such as
the Internet begins with a slower rate of adoption, but then has a more "pronounced"
critical mass effect; once the critical mass is reached, the innovation takes off more
rapidly than projected by the traditional S-curve (Rogers, 1995; Mahler & Rogers, 1999;
Garrison, 2000). In other words, the adoption of interactive innovations follows a
The importance of the (perceived) number of other adopters of the interactive
technology decreases over time. Today fax machines and telephones are so widespread
(at least in the United States) that potential adopters don't have to consider the interactive
nature of these communication technologies. As Mahler and Rogers (1999, 720) point out
about telephone adoption today, "the utility of adopting depends almost entirely on
factors internal to the individual, rather than on such externalities as the perceived
proportion of others with whom the individual wishes to communicate by telephone."
The Internet in the United States has probably reached that point. If we study Internet
adoption in a developing country, however, network externalities are likely to affect the
Internet diffusion process (Maherzi, 1997).
Network externalities are often cited as attributes of information goods and
services. Basically, network externalities add value to a certain product or service with
the increase in the number of its users. Shapiro and Varian (1999) discuss in economic
terms how information goods have higher network externalities and thus are more
valuable to both the user and the provider. In the case of the Internet— the so-called global
network of networks, externalities stemming from the number of others who have already
adopted it should be highly visible. As the Internet becomes more and more prevalent,
and especially after a critical mass is reached, the significance of the total number of
users will fade away.
Network externalities can be direct or indirect. As Mahler and Rogers (1999) note,
there were indirect network externalities in the case of the VCR diffusion. Two different
standards were developed—Beta and VHS; tapes dubbed in one of these standards
couldn't be played on the other. Yet the VCR standard that was adopted was VHS, even
though it was the technologically-inferior one (this was largely due to better marketing
efforts on the part of VHS). In the VCR case, critical mass had to be reached for each of
the two standards (Mahler & Rogers, 1999). This example is applicable to Internet
adoption, and especially to the adoption of specific Internet software. The diffusion of
online chat programs such as ICQ will depend largely on the ICQ company's efforts to
make their program the standard in online chatting.
One way to study the Internet using traditional diffusion studies is by positioning
it as an interactive innovation. Another approach is to examine it as a cluster innovation.
Prescott and Slyke (1997) argue that the Internet can be best understood as a cluster
innovation because of its many components. Clearly, classification of the Internet as an
innovation is hard because of its very complex nature. As Prescott and Slyke (1997)
suggested, a good way to understand the Internet may be by positioning it as a technology
cluster innovation— i.e., not as one single technology, but as several technologies working
with one another. The Internet has many components, many of which are related to or
dependent on each other. If people adopt email technology, they are more likely to adopt
Instant Messenger, for example. They will not be able to do this, however, unless they
have the Internet Explorer browser installed already.
If we see the Internet as a cluster innovation (several technologies working
together), we see that it has many relative advantages, depending on which purpose you
use it for. The Internet is not compatible with anything that existed prior to it, its
networked structure and way of condensing space and time have been unprecedented.
The only other technological innovation in history that had such dramatic effect was
probably the telegraph-"shrinking the world faster and further than ever before"
(Standage, 1999, vii).
Both cluster innovations and interactive innovations will have different degrees of
complexity, trialability and observability. The Internet today may be easier to try out
compared to five years ago. Any American can go to the public library or to a friend's
house, for example, and try using the Internet or any of its components and/or services.
Its results are also readily observable. In general, that should contribute to a faster rate of
When we talk about the diffusion of an innovation, we basically examine behavior
change over time. In this case, we are interested in how many people have adopted the
new technology within a country. Thus we are interested in measuring the number of
adopters who have reached the implementation stage in the adoption process. It will be
useful to review next what type of adopter categories exist in diffusion literature.
Types of Adopters
There are five basic adopter categories described in diffusion of innovations
studies on the basis of how fast a member of the social system adopts a new idea, product
or technology. They are: (1) innovators, (2) early adopters, (3) early majority, (4) late
majority, and (5) laggards (Rogers, 1995). Diffusion of innovations posits that innovators
are venturesome, cosmopolite, with higher income and technical knowledge (Rogers,
1986; Rogers, 1995). The innovators account for 2.5 percent of the total population that
in the end adopts the innovation. Typically, the early adopters are about 13.5 percent, the
early majority and the late majority groups are 34 percent of the population each, and the
laggards account for 16 percent (Rogers, 1995). This reflects the S-curve adoption
process and follows from the normal distribution of the total number of new technology
Using Rogers' normal distribution curve of innovators, early adopters, early
majority, late majority, and laggards, it is easy to predict in which category future
Internet adopters will fall. This is a useful technique when examining adoption at the
Internet User Profile
An increasing number of research studies have been conducted in the United
States to discover the "profile" of Internet users. A number of user studies conducted by
the Georgia Tech GVU lab in the past years show that most Internet users in the United
States are male, well educated, and upper income (GVU, 1999). in a study of American
students Stewart et al. (1998) conclude that, in general, men are more willing to adopt a
new technology than women. Interestingly, they also suggest that whites are more willing
to adopt than other cultural groups in the United States.
Lindstrom (1997) finds similar characteristics for North American Internet users:
they are mostly male, upper class and well educated. Lindstrom' s survey looks at users in
the United States and in Canada during two time periods. The second survey shows that
the typical user has changed and that the user base has broadened (Lindstrom, 1997). The
typical Internet user in the United States has changed. More women and older people
have joined the online community. In fact, more recent statistics show that the majority of
American Internet users are women (Nielsen NetRatings, 2000). This remained true in
2002 (Nielsen NetRatings, 2002).
What is the Internet user profile in other countries? Research on Internet use in
Chile, for example, shows some similarities: Internet users in the city of Santiago
resemble closely early American users (Mendoza & Alvarez de Toledo, 1997). The
typical Chilean Internet user can be characterized as young, male, and highly educated.
The study also reveals that Chilean users tend to have higher income level and to connect
to the Internet from work or from educational institutions (Mendoza & Alvarez de
One of the few studies on Internet users in Eastern Europe shows a similar
typology (Dimitrova, 2002). Dimitrova's survey of Bulgarian Internet users finds that the
majority are highly educated and male (Dimitrova, 2002). Her questionnaire does not
include income level questions. Similarities in user characteristics are also found in Asian
countries. The typical Internet user in China, for instance, is also more educated, young,
and mostly male (Jisi et al., 2001). In addition to the United States, Canada is the only
other country where the number of female Internet users has surpassed that of male users
(Nielsen NetRatings, 2002).
Many studies have tried to explain why some people have adopted the Internet
while others have not (Goode & Stevens, 2000). Atkin, Jeffres, and Neuendorf (1998)
examine what differences exist between Internet adopters and non-adopters, and whether
those individual characteristics can be used to predict the potential users of Internet
technology. Atkin, Jeffres, and Neuendorf (1998) draw from several theoretical
frameworks. They see the computer as one of the most discontinuous innovations ever.
They include media usage variables when looking at what characteristics differentiate
early Internet adopters from late adopters.
Atkin, Jeffres, and Neuendorf (1998) examine the demographic characteristics of
the Internet adopters and non-adopters to test the diffusion of innovations paradigm. As
suggested in the diffusion typology, they find that Internet users tend to be better-off and
better educated than the population at large. Atkin, Jeffres, and Neuendorf (1998) also
find that prior use of and interest in technology is a good predictor of Internet adoption,
which also fits with Rogers's technologically-knowledgeable innovator profile. Internet
adopters are also found to be more cosmopolite than non-adopters (Atkin et. al, 1998).
Many studies support Rogers' generalization that early Internet adopters are better
educated and upscale (Atkin et. al, 1998; Lin, 1998). Interestingly, several studies suggest
that individual demographic factors are statistically more significant in predicting Internet
adoption than attitudinal or communication needs factors (Atkin et. al, 1998). Media
usage is not an important predictor of computer and Internet adoption (Atkin et. al, 1998;
GVU, 1998; ITU, 1999; Lin, 1998). Demographic segmentation then can be used by
diffusion agencies to target specific adopter groups in the diffusion of new technologies,
especially in the early stages of adoption. Demographic characteristics also should be
used as predictors when studying Internet adoption by countries with lower Internet
Other Communication Technologies
Garrison (2000) looks at the diffusion of online research tools in American
newsrooms. These tools represent a particular aspect of the Internet and its diffusion to a
particular group of users— American journalists. Garrison (2000) concludes that
computers are already largely adopted as a newsgathering method and that the "value of
interactive Internet information-gathering tools" has increased over time.
The rate of adoption (the number of people who adopt an innovation over a
specified period of time) varies greatly between different technological innovations. The
adoption of Nintendo games in the United States, for example, and the VCR was very
fast (Rogers, 1986). Computers, however, have exhibited a slower rate of adoption. Weir
(1999) examines the adoption of electronic newspapers and finds that it is different from
that of other consumer products. Weir (1999) agues that this may be due to the fact that
electronic newspapers are more like media than like software applications.
We cannot observe Internet diffusion unless there are personal computers adopted
in the first place. As Lin (1998) indicates, PC's adoption rate in the United States has
been relatively slow. She also finds that communication technology ownership is the
strongest predictor of computer adoption rate (Lin, 1998). Lin argues that demographic
predictors are still important to forecast computer adoption rates, but that is no longer the
case for the VCR and cable television. This suggests that communication technologies'
adoption and diffusion changes over time, both in terms of speed and type of adopter.
This, in fact, supports Rogers's model, in which it is expected that innovators and early
adopters may be demographically different than the later adopter groups. Among the
laggards, early majority, and late majority there are no major demographic distinctions.
Therefore, the strength of demographic predictors decreases over time.
Prescott and Slyke (1997) argue that (1) the Internet is a complex technology
cluster and (2) its adoption is context-specific. The researchers ask the following
questions regarding Internet adoption at the organizational level: Is the Internet radical
versus incremental innovation, product versus process innovation, voluntary versus
involuntary adoption, and pull versus push technology. Internet adoption will have
different dimensions in different contexts. For example, when a company adopts new
software that does not change significantly the way business is conducted (just changes
the application from text-based to windows-based software, for instance), this is an
example of an incremental change (Prescott & Slyke, 1997). On the other hand, if the
company introduces a Web-based software that allows customers for the first time to
place orders online, we have a radical shift from previous organizational practices
(Prescott & Slyke, 1997). The type of Internet adoption is contextual then-on a case by
case basis, Internet adoption can fall into one or the other category.
Levels of Internet Adoption
Not only is Internet adoption contextual, but it can also be divided into three
different levels: individual, organizational, and societal. Most research to date has
focused on either individual or organizational adoption. As Rogers (1995) says, however,
we can also study adoption at the societal level.
One of the few studies that applies diffusion of innovations to the societal level
conducted by Bazar and Boalch (1997) supports the traditional diffusion of innovations
typology. They argue that Internet diffusion in developing countries is achieved when a
critical mass is reached and the adoption becomes self-sustaining (Bazar & Boalch,
1997). The main institutions that play a role in the Internet adoption process within a
country, according to Bazar and Boalch (1997), are the government, the carriers, funding
institutions, and the information technology (IT) professional associations. Others have
also found that national governments affect significantly the Internet diffusion process
There are three different ways of approaching Internet adoption in terms of which
level of adoption we are looking at. These levels, again, are the individual,
organizational, and societal level. With a global innovation such as the Internet, however,
we need more insights about the societal level of adoption. At this level, different factors
affect the rate of adoption, and diffusion of innovations, while useful, is not sufficient to
explain country-level technology diffusion. A major limitation of the diffusion of
innovations typology is the fact that it has been applied mainly to the individual and
organizational levels. This dissertation contributes to knowledge on diffusion of
innovations at the societal level.
Other Considerations for Internet Adoption
Few studies on Internet adoption make a clear distinction between the software and
hardware parts of the Internet, a distinction that Rogers (1995) considers important for
any technology. Such distinction may reveal different patterns of adoption. In addition,
we need to consider which aspect of the Internet is being examined (e.g., email) when we
look at the adoption of the Internet from a diffusion of innovations perspective. Such
delimitations are important yet difficult to make.
Internet adoption is also subject to whether it is voluntary or not for the individual
adopter. As Rogers (1995) points out, there are authority or contingent decisions that
apply to certain adoption situations. The diffusion process is different in cases of optional
or collective adoption decisions. In this study, we assume that Internet users in the post-
communist world have made a voluntary decision to use the Internet.
Wolcott et al. (2001) incorporate national systems of innovation (NSI) literature in
their discussion of Internet diffusion theory. This body of research underscores the
importance of national institutions for the diffusion of innovations within countries. NSI
studies often include Research and Development (R&D) expenditure as a factor affecting
the successful adoption of the innovation as well as other measures of training and
education (Nelson, 1993). Typically, countries with higher income levels have higher
R&D expenditure so including both variables as determinants of Internet usage rates
could be redundant.
A better conceptualization of the Internet itself is needed before we can fully
understand its adoption at either the individual, organizational, or societal level. Some see
the Internet as a technological innovation. Others describe it as a cluster of technologies.
Some see the Internet as a culture (Wilson et al., 1996). Yet others talk about the Internet
as a place (Poster, 1995). Many portray the Internet as a strategic national infrastructure
(Madon, 2000). Others have discussed its potential to enhance democratic governance
and improve socio-economic development of developing countries in particular
The Internet is very complex in nature and is still evolving. Therefore, further
research on Internet diffusion and adoption is needed. The diffusion of innovations
typology is useful overall, but cannot fully explain the adoption at the country level. This
dissertation also draws on new media technologies literature, which is summarized in the
New Media Technologies Research
Diffusion of innovations basically postulates that the diffusion process occurs in
stages over time. A main assumption of that theoretical paradigm is that (1) the diffusion
process is linear and (2) different groups adopt innovations at different points in time.
The width of adoption that can describe how widespread a technology is within a country
is largely based on individual factors. Thus, audience variables are critical for explaining
levels of Internet adoption. In particular, demographic factors such as personal income,
and educational level, and attitudinal factors such as cosmopoliteness and innovativeness
have been identified as important in the diffusion of innovations model.
According to diffusion of innovations, external factors are also important for
Internet adoption in developing countries. These include government policies and the
existence of prior technologies. Multiple technology components are critical in the case
of the Internet, as the above discussion of interactive and cluster innovations shows.
Therefore, government policies and political environment as well as technological
infrastructure can be used as predictors for Internet adoption.
The nature of interactive communication technologies and the Internet in particular
requires us to draw from a multidisciplinary theoretical framework to explain variations
in adoption (Lin, 1998). Several basic sets of factors have emerged in the growing body
of new media technologies literature as predictors of country level Internet adoption.
These are economic factors, political climate and policy factors, technology and
infrastructure, audience characteristics, and cultural factors. This study proposes a
conceptual framework including the afore mentioned five areas to explain Internet
adoption at the societal level.
While certain prerequisites exist that affect adoption of new technologies, it is a
difficult task to pinpoint exactly what drives Internet diffusion into different countries.
Most studies to date have shown that the country's economic development plays an
important role. The next section reviews literature on Internet adoption and economic
The argument exists that innovations are borne as a result of scientific discoveries.
Some scholars have argued against this proposition, however, saying that the rate of
technological progress does not stem directly from basic scientific discovering (Romer,
1999). Rather, "it is the incentives created by the market that profoundly affect the pace
and direction of economic progress" (Romer, 1999). Logically, then, technology growth
is faster when these incentives are stronger. Many scholars have argued that business
incentives and potential markets determine whether a new technology is introduced or not
in the first place (Romer, 1999; Sale, 1999). Malecki (2001), for example, contends that
commercialization has been the main catalyst for the development and growth of Internet
The most evident predictor of Internet penetration in a country is probably the level
of economic development (Arnum & Conti, 1998; Bazar & Boalch, 1997; Elie, 1998;
Hargittai, 1999; Wolcott et al., 2001). The country's economic situation has a direct
effect on Internet adoption. Studies have consistently detected a strong positive
correlation between the level of economic development and Internet use in a country
(Arnum & Conti, 1998; Clarke, 2001; Elie, 1998; Gunarante, 2001; Hargittai, 1999;
Kiiski and Pohjola, 2001). Using World Bank data and classification, Figure 3 shows the
uneven distribution of Internet hosts across countries with different income levels.
In addition to empirical data, previous research also shows that countries that are
better off economically tend to have higher Internet penetration. Bazar and Boalch (1997)
contend that capital (economic resources) are crucial for adoption of the Internet in
developing countries. Richer countries in general have more resources to put into the
service sector of their economies, which includes the information technologies (IT) sector
(Elie, 1998). Previous research has also supported the assumption that richer countries
have more telecommunications networks and higher media penetration overall (Maherzi,
Internet Hosts by Income Regions
Lower middle income Upper middle income High income: nonOECD High income: OECD
Figure 3-1. Internet hosts across income regions. Source: World Bank, 2000.
Bazar and Boalch (1997) examine Internet diffusion within developing countries'
context. They define five categories that can be seen as prerequisites for Internet adoption
in developing countries: national/organizational needs and opportunities in place (this
concept is related to the nation's vision of the future and relative advantage of the
country); technology (including infrastructure and Internet technology itself); necessary
skills and people to introduce the innovation; capital (economic resources); and finally
good management of the technology adoption and diffusion process. Again, they contend
that economic resources are critical for Internet adoption at the societal level (Bazar &
One of the few studies on Internet growth in the post-communist countries focuses
on Internet use in enterprises within the countries (Clarke, 2001). Thus, the unit of
analysis is the individual enterprise and the dependent variable in the econometric model
is whether or not an enterprise has access to the Internet. The main question that the study
addresses is how enterprise ownership and foreign competition affect Internet access in
the region. The sample includes 2,999 enterprises from Eastern Europe and Central Asia
and uses the WBES database of the World Bank (Clarke, 2001). The main conclusion of
the study is that foreign ownership of enterprises positively affects Internet growth. In
addition, the results suggest that enterprises in smaller countries with higher income
levels and larger urban populations are more likely to be connected to the Internet
Most studies of cross-country Internet adoption to date show that national income
level is an important determinant. Elie (1998) and Hargittai (1999) find a strong
correlation between Internet penetration in a country and per capita income. Hargittai
(1999) examines how four country-level indicators affect Internet connectivity among
OECD members. She includes the economic situation of the country (measured by GDP
per capita) as well as education level, legal environment (regarding communication
technologies) as well as infrastructure to explain differences in Internet connectivity. The
GDP per capita is the strongest predictor in her model: it explains 38 percent of the
variation. Rodriguez and Wilson (2000) also use GDP per capita as a predictor of ICTs
use, as measured by and Index of Technological Progress (ITP). They find a strong
positive correlation between ITP and GNP PPP, also arguing that richer countries make
more technological progress over time than poorer countries.
Clearly, GDP or other macroeconomic indicators can be used as predictors of
Internet adoption in a country. This is also a logical variable since it translates individual
income, which has been identified as important in the diffusion of innovations paradigm,
to the country level.
Kiiski and Pohjola (2001) look at cross-country diffusion of the Internet using the
Gompertz model of technology diffusion. Thus, their model is longitudinal and their
dependent variable is the rate of change in the number of Internet hosts from 1995 to
2000. Their sample includes 23 OECD countries, similarly to Hargittai (1999). Their
study concludes that the best predictors for adoption are GDP per capita and Internet
Some researchers, however, find no relationship between national income and
Internet penetration. Surprisingly, the parameter for income per capita is statistically
insignificant in a recent study at the World Bank (Dasgupta et al., 2001). Thus, the
researchers conclude that economic development does not have a strong influence on
Internet intensity. They extend this conclusion by noting that the disparity in Internet use
is just a reflection of the "long-standing disparity in telecommunications access" between
developed and developing countries (Dasgupta et al., 2001, 6). One possible explanation
for this finding is that income and teledensity (telephone penetration per capita) are
highly correlated and when one of them is included, the other does not appear to have an
effect. However, there is no information on the bivariate correlations among the
predictors given in the study.
Another economic variable that is clearly related to Internet use is price of Internet
access. Petrazzini and Guerrero (2000) argue that there is an inverse relationship between
price of Internet connection and Internet use. They find that once the price of leased lines
and of tariffs for local calls in Argentina has been reduced, Internet growth increases
dramatically. Higher Internet prices are generally indicative of more restrictive
telecommunications policies. They also present a barrier to Internet use of lower income
demographic groups. However, the price of Internet access in the post-communist
countries is not available; neither is any other synchronized country-level data on
governmental policies regarding Internet access.
In addition to price of Internet connection and basic macroeconomic indicators, the
size of the service sector of the economy has also been found an important factor for IT
diffusion (Elie, 1998). The size of the service sector of the economy, however, is
typically related to national income. Countries with higher per capita income tend to have
bigger service sectors (World Bank, 2000).
When comparing some Western European and some Eastern European countries,
Elie (1998) finds that Internet penetration differs from what would be expected only on
the basis of macroeconomic indicators, however. Economically highly developed
countries of Southern and Central Europe (such as France, Germany, Italy, and Spain)
have lower Internet usage levels than predicted on the basis of their GDP levels (Forrester
Research, 2000). On the contrary, the Internet seems more "developed" in the
economically less advanced Eastern European countries of Slovenia, Czech Republic,
Hungary, Slovakia, and Poland.
Arnum and Conti (1998) note a similar discrepancy. When calculating Internet
ratio per country, they show that France and Estonia are very close to each other in their
Internet penetration. Similarly, Estonia and Slovenia were ranked higher than Hong
" Instead of using the conventional measures of Internet hosts or Internet users per 10,000 people, Arnum
and Conti measured Internet penetration per country as what they called an Internet ratio. The Internet ratio
formula used in their study is: (Internet Hosts + Domains + Web Pages)/ Population.
Kong, Portugal, and Greece, in terms of their Internet ratio (Arnum & Conti, 1998). The
study also concludes that several countries with relatively high levels of economic
activity—for example, Saudi Arabia, Oman, Venezuela, and the U.A.E.— have surprisingly
low levels of Internet activity (Arnum & Conti, 1998). These results suggest that
although economic factors are important, there is more at stake in the case of Internet
adoption at the societal level.
Political Climate and Policy
Economic indicators by themselves cannot fully explain the diffusion of interactive
communication technologies, such as the Internet. Even among rich countries, there is a
large amount of variation in Internet penetration (ITU, 2000). A study focusing on the
Organisation for Economic Cooperation and Development (OECD) members concludes
that GDP by itself explains less than 40 percent of the variation in Internet connectivity
(Hargittai, 1999). Other studies have indicated the importance of the political stability of
a country on its overall development and adoption of new technologies (Berg-Schlosser
& Siegler, 1990). Clearly, political instability will present an obstacle to fast Internet
diffusion. A good example to support this claim is Rwanda, a country with serious
political conflict in recent years. The most recent data from the World Development
Report of the World Bank show that Rwanda has practically zero Internet usage (World
The democratization level of the country has been suggested to be an important
predictor of Internet usage. Studies show some evidence that political freedoms are
positively related to Internet use (Daly, 2000; Norris, 2001; Rodriguez & Wilson, 2000).
Rodriguez and Wilson (2000), for instance, argue that a national democratic system is
critical in the adoption of information technologies such as the Internet. They underscore
the significance of democratic rights and civil liberties for the creation of a climate where
information and communication technologies (ICTs) can be easily adopted (Rodriguez &
Wilson, 2000). Most accounts show a reverse relationship: countries with restricted
political and civil liberties tend to have lower Internet usage levels, while democratic
societies tend to encourage Internet growth.
Rodriguez and Wilson (2000) measure technological progress as a combination of
TV sets, mobile phones, personal computers, Internet hosts, and fax machines. Their
study shows the following four factors are critical for national technological progress: a
climate of democratic freedoms that facilitates the adoption of ICTs; rule of law and security
of property rights; investment in human capital; and finally low levels of government
distortions. Interestingly, they found that a transition from the least free stage to a higher
stage of civil liberties led to an increase in growth rate of technology of 1 8 percentage
points (Rodriguez & Wilson, 2000). They also claim that "developing countries that
successfully innovate and diffuse ICTs are able to open their political systems as well as
their investment and commercial institutions" (Rodriguez & Wilson, 2000, 28). It seems
that the relationship between Internet use and political freedoms may be circular.
Another study testing the relationship between democracy and Internet use shows
that Internet penetration was highly correlated with higher political freedoms and
democratization levels (Norris, 2001). Norris (2001) also argues that Internet adoption
within a national context is generally affected by two broad factors: (1) socioeconomic
development and (2) democratization. Empirical testing showed that national level of
democratization is highly correlated with a New Media Index (Norris, 2001). When
controlling for income levels, however, democratization becomes insignificant (Norris,
One way of measuring democratization is by looking at the level of civil liberties
(Pritchett & Kaufmann, 1998; Rodriguez & Wilson, 2000). Definitions of civil liberties
vary. One organization that has consistently studied the level of civil liberties around the
globe is the Freedom House. Their civil liberties ranking focuses on several areas,
including freedom of expression and belief, independent media, freedom of assembly and
demonstration, rule of law and human rights, and personal autonomy and economic rights
in the country. The Freedom House civil liberties ranking emerges as the best proxy for
democratization and will be used as a predictor in this study.
In addition to democratization, government effectiveness can also directly impact
the rate of adoption of new technologies such as the Internet. Previous studies have
shown that the role of government is crucial in the early stages of Internet adoption
within a country (Bazar & Boalch, 1997; Lin, 1993). Lin (1993) argues that the American
government, for example, has played a critical role in the development and growth of the
Internet in the United States. National government policy has been critical for Internet
growth in Latin America and Western Europe as well (Petrazzini & Guerrero, 2000;
Policies are directly related to the political climate in a country (Godwin, 1998). A
study of cross-country Internet diffusion includes policy and urbanization variables in
addition to income to explain diffusion levels (Dasgupta, et al., 2001). Hargittai (1999)
finds a strong explanatory power in policies regarding the telecommunication sector in
her study of Internet connectivity among the OECD countries. Those countries that have
allowed free competition or even some degree of competition have a higher level of
Internet penetration than countries with telecom monopolies, other things being equal
Kiiski and Pohjola (2001) also include competition in telecommunications
(measured as existence of some form of competition in telecom markets) as a predictor of
Internet use. However, competition in their model is not found to be significant, in
contrast to Hargittai (1999). One possible explanation is that they use both competition
and access prices in one regression model. That combination can be redundant as the two
variables are likely to be correlated. Typically there will be an inverse relationship: more
competition will bring lower prices and vice versa.
Other researchers have also claimed that it is critical to take into account
government policies when trying to examine levels of Internet adoption (Petrazzini &
Guerrero, 2000; Sallai, 2000; Wolcott et al., 2001). This is especially important in the
post-communist countries, which have been undergoing major political transformation
have been trying to open up their telecommunications markets only since the early 1990s.
There is no doubt that government policies impact the pace of new technology
adoption. Changes in the telecommunications industry in the 1970s and in the 1980s in
particular have resulted in the introduction of privatization and liberalization policies in
many countries around the world (Bauer, 1994; Gruber, 2001). Western European nations
undertook a major shift toward telecom liberalization in the 1980s. Bauer argues that
three main factors contributed to this shift: innovative equipment resulting from rapid
technological changes; demand for customized and specialized telecommunications
services; and desire by the large telecom equipment providers to enter foreign markets.
It is important to understand the significance of national policy for the creation of
supportive Internet environment. Specifically, government support is needed to make the
Internet affordable for the population at large (The Internet's new borders, 2001). A
competitive telecommunications market becomes critical not only for the development of
e-business, but for making the Internet more accessible to people. A World Bank report
(1999) finds that telephone networks expand much faster in those countries that have
privatized their telecommunications market. Even if the telecom operator is a privatized
monopoly, it is claimed to be better than a state monopoly operator. Clearly national
policies in the telecommunications sector are important and need to be considered in the
study of country-level Internet adoption.
Government policy is found significant in a study of national computer imports.
Caselli and Coleman (2001) examine cross-country technology diffusion by looking at
the determinants of computer imports. Their study suggests that computer adoption is
related to policy: higher levels of trade openness towards the OECD positively affect the
number of computers in this case. The results also show that income per worker,
investment per worker, and secondary education are significant predictors. All of these
factors are positively related to computer adoption. The study also finds that computer
adoption is negatively affected when having a large government share in GDP as well as
large share of agriculture in GDP (Caselli & Coleman, 2001).
Privatization and liberalization policies directly affect the price of
telecommunications devices and services. Prices have been claimed to be higher in
countries with monopolistic telecommunications markets (Horvath, 2002; Jamison, 1995;
Ryan, 1997). Fish (1998) underscores the importance of privatization and liberalization in
the post-communist world in explaining, at least in part, the relative differences across
these countries (Fish, 1998).
Previous research shows that in general there are several stages in
telecommunications reform. At the first stage typically we observe privatization of the
incumbent state-owned telecom operator. At the second stage, competition in the telecom
market is introduced. General, it has been found that competition is more beneficial to the
consumer of telecommunications services.
Fish (1998) examines the power of privatization and liberalization combined as
determinants of the long-term economic reform in the post-communist countries. These
two sets of policies are important not only for the success of the economic reform in the
region, but also for the facilitation of telecommunications development and restructuring
from a centralized model to an open-market model. Paltridge (2000) argues that
liberalization in telecommunications facilitates Internet development. Internet access
prices are important factors as suggested by the vast variability in Internet usage across
the OECD group (Paltridge, 2000).
Sometimes privatization needs to be preceded by the adoption of a clear-cut
regulatory framework in the country (Wheatley, 1999). Wallsten (2002) argues that
institutional reform and regulations need to be in place before the telecommunications
firm is privatized in order to have a positive effect. In other words, he recommends
having regulation first and then subsequent privatization in the telecommunications sector
in any country.
It is critical to include the level of privatization when studying Internet penetration
in the post-communist countries (Estache et al., 2002, Kuentzel et al., 2000; Maddock,
1997). Privatization has been a difficult process in the transition from centralized state
economies to market-based economies in those countries (Ellis, 1999; Fish, 1998;
Gospic, et al., 2000; Gulyas, 1998; Hoelschner, 2000; Jasinski, 1997; Papir & Oleszak,
2000; UNDP, 1999). As Bauer argues (1994), the post-communist countries "often have
to create very favorable conditions for infrastructure service providers ... to succeed in
the attraction of foreign investment capital and technology." Lari (2000) underscores the
importance of both financial and non-financial aid that international institutions need to
provide for Eastern European countries. This need is especially acute in the area of
telecommunications (Lari, 2000). In addition to attracting foreign capital, privatization in
the telecommunications sector is also very important for competition and Internet growth
in general (Gulyas, 1998; Sallai, 2000; Sokolov & Goldenstein, 2000).
As Maddock (1997, 166) points out, "Eastern Europe has followed the Latin
American model by relying on liberalization to achieve reform but social and political
disruption has meant that the potential gains have not yet been achieved." In all post-
communist countries, the main telecommunications operator used to have a strong
monopoly in the domestic market. In the past, the incumbent telecommunications
operator often served as a tax collector and regulator (Canning, 1997; Maddock, 1997;
Michalis & Takla, 1997; Xavier, 2000). The pricing structure was distorted by cross
subsidies where domestic calls were artificially kept at a lower price, which was
compensated for most often by high international tariffs.
Campbell (1995), in his overview of the telecommunications industry in the former
Soviet Union, makes several observations, which are generally true for the rest of the
Soviet bloc. He notes that the control of the Post, Telephone, & Telegraph (PTT) is
completely in the hands of the government. The monopoly, he notes, will be rather
difficult to break as this sector is considered strategically important (Campbell, 1995).
The Communist leaders did not view information and communications networks as its
top priority. Campbell (1995, 25) concludes: "The telecoms sector illustrates the general
problem that it is not that easy to just start over. The old system left a technological and
organizational legacy that cannot be overcome quickly." The old structures— both
institutional and physical— will take time to improve and open up even though the Cold
War is over.
The telecommunications sector has being undergoing major reforms since then, but
it has proven to be quite difficult to modernize. Looking at the former Soviet republics,
Campbell notes that "telecoms policy is a result of what is happening under the general
processes of privatization, antimonopoly policy, price regulation, and tax policy as much
as by specific legislation on telecommunications" (Campbell, 1995, 207). Among the
former Soviet republics, the Baltic states have been most liberal in giving autonomy to
the telecom operator. Ukraine and Belarus, on the other hand, show less desire to change
quickly. The reforms in these countries have been slower (Campbell, 1995). Central and
Eastern European countries are trying to get closer to the regulatory framework of the
European Union with their telecom legislation. The process of liberalization of national
telecom markets is going faster in these five countries: the Czech Republic, Estonia,
Hungary, Poland, and Slovenia (Bruce, 1999). However, there are different degrees of
progress not only across the region, but also across these countries.
Dasgupta et al. (2001) show the importance of policy, measured as the degree of
private sector competition in a study of cross-country Internet diffusion. However, this
measure does not directly capture the level of openness in the telecommunications sector
in particular. Thus, the policy variable in this study can be questionable, even though it
shows the expected direction of impact on Internet use. In addition, the policy index used
in the study is based on data from 1995 while the rest of the variables are from 1990. This
is another shortcoming of the policy predictor. This is a case in point that it is very
challenging to find a good measure of national policies, especially as they relate to
Indeed, it is difficult to measure the recent liberalization and privatization policies
in the post-communist countries. It can be argued that the overall level of market
openness within a country is reflected by the level of economic freedom, so the
Economic Freedom index of the Heritage Foundation could have been used. However, as
noted about the Dasgupta et al. study (2001), this can be misleading because if a country
has adopted liberal economic policies overall, its policies in the telecommunications
sector could still be very restrictive. Thus the level of privatization in telecommunications
specifically needs to be used, when data permit, to explain Internet adoption levels.
Privatization by itself, however, may lead to negative results in some cases.
Wallsten (1999) examined the effects of competition, privatization, and regulation in the
telecom markets in a number of African and Latin American countries. He found that
when regulation is introduced with the privatization reform, the effects tend to be
positive. Again, regulation is also an important aspect of telecommunications reform.
This study, however, measures only one aspect of telecommunications reform— telecom
privatization, which is only the first stage in the liberalization process. Future studies
should try to incorporate not only privatization, but also completion and regulation
variables, whenever possible.
Hargittai (1999, 705) notes that "existing telecommunication facilities may be
crucial for understanding variation in the spread of the Internet." Bazar and Boalch
(1997) identify technology as one of five determinants Internet adoption in developing
countries, in addition to economic resources, needs and opportunities, necessary skills
and people to introduce the innovation, and finally good management of the diffusion
process. They position technology to include not only Internet technology itself, but the
general infrastructure within the country (Bazar & Boalch, 1997).
Arnum and Conti (1998) argue that "[w]hat TV was to the second half of this
century, what the telephone and the paved road were to the early 20 th century, and what
the railroad was to the 19 th century, so too is the Internet to the current generation."
Arnum and Conti (1998) relate the speed of adoption of new technologies to the prior
existence of other infrastructure. They conclude that the Internet is more popular in
countries that have long-established infrastructures for communications and
transportation. Western European countries then will be more likely to have higher
Internet adoption rates, as they have historically had widespread networks of
transportation, communications, and other technological infrastructure.
Daly (1999) emphasizes that infrastructure in general varies widely across regions
of the world. Since existing infrastructure significantly affects Internet adoption in a
country (Arnum & Conti, 1998; Bazar & Boalch, 1997; Elie, 1998; Gulyas, 1998;
Hargittai, 1999; Lin, 1998; Sadowsky, 1993), variations in Internet use are to be
expected. Gulyas (1998) argues that a modern telecommunications network is a basic
requirement for a society to become an information society. Elie (1998) contends that the
existence of a telecommunications network is critical for Internet adoption. Sadowsky
(1993) discusses the substantial physical and capital investment required to build Internet
infrastructure before a country can benefit from the Internet. He adds that the
infrastructure in many developing countries is inadequate for more advanced network
activities (Sadowsky, 1993).
In the former Soviet bloc, telephone infrastructure is generally inferior than the one
in place in Western European countries. Looking at the ex-Soviet republics, Cambell
(1995) notes that the telephone infrastructure in those countries is outdated and, therefore,
the quality of service is low. He says: "Old-fashioned and worn-out switching equipment
meant bad connections" (Cambell, 1995, 25).
Kiiski and Pohjola's study (2001) shows that infrastructure variables are critical for
the increase in the number of Internet hosts per capita. Specifically, per capita telephone
lines and number of PC's are included in their estimation (Kiiski & Pohjola, 2001). The
researchers note that both telephone lines and PC's are strongly related to GDP so they
may have a somewhat indirect effect on cross-country Internet diffusion. However, no
partial correlations are provided.
One way to measure the existing telecommunications infrastructure in a country is
telephone penetration (teledensity). Telephone infrastructure has been traditionally used
in studies on Internet penetration (Clarke, 2001; Elie, 1998; Guillen & Suarez, 2001;
Hargittai, 1999; Kiiski & Pohjola, 2001). The analytical framework proposed here will
use telephone density as an infrastructure indicator, but will also include mobile phones
in addition to residential phones.
Jupiter Research (2001b) projects a significant global increase in the number of
mobile Internet users. In the developing world in general, mobile phones seem to play an
important role. In some African countries, the number of mobile phone subscribers
surpasses those that use residential phones. Interestingly, Uganda has more mobile phone
customers than fixed telephone customers as of July 1999 (Minges, 2001). The popularity
of wireless is evident in the Latin American region as well. Research shows that there is
an emerging audience in Latin America which will access the Internet only or primarily
from a mobile phone (Jupiter Research, 2001). The growth of the mobile telephone
market in Eastern European countries has served as a major incentive for
demonopolization of telecommunications service overall (Oaca, 2000).
It is true that Internet usage is dependent on phone line availability and also on
computer availability. Clearly, to connect to the World Wide Web one needs to have a
personal computer in the first place. Yet lack of data prevent us from using number of
computers per capita as an independent variable in this study. Also, even though the
association between Internet users and number of computers is relatively clear, the
direction of causality between the two variables is not so obvious.
Another question that may arise is why not include cable infrastructure to account
for broadband Internet users. Broadband use in the post-communist countries is very
limited. Even in Western Europe, broadband Internet has not become very popular
(Pastore, 2002). Pastore notes that Western European countries are not yet ready to go
broadband. Specifically, the number of broadband users of all Internet households is
about eight percent in France, nine in Germany and a very low two percent in Britain
(Pastore, 2002). In the post-communist countries, it will be quite a while before
broadband technology becomes available nationwide.
Research to date has focused more on the supply side rather than the demand side
when examining Internet diffusion within countries (Lamberton, 1997). Researchers have
attempted to answer the question what demographic and attitudinal characteristics affect
Internet adoption. Diffusion of innovations (reviewed above) shows that innovators and
early adopters typically have higher socio-economic status, are better educated, more
cosmopolite, and technologically savvy (Atkin et al., 1998; Rogers, 1995).
Audience characteristics affect Internet adoption in several ways then. First of all,
education has emerged as a major determinant of Internet adoption both at the individual
and country level. The higher the education level of the general population, the more
likely people are to adopt new media technologies such as the Internet (Caselli &
Coleman, 2000; Hargittai, 1999; World Bank, 1999). Lack of adequate education, on the
other hand, can impede Internet diffusion. El-Nawawy (2000), for example, sees
education as the primary deterrent to Internet growth in the case of Internet adoption in
Caselli and Coleman (2000) contend that the choice of technology is driven by the
human capital in a country. They conclude that countries that have more skilled labor
adopt technologies that efficiently use that labor, which in turn leads to more capital. In
contrast, countries with more unskilled labor adopt "less sophisticated" technologies and
accumulate less capital as a result (Caselli and Coleman, 2000). Press et al. (1998) note
that there are several determinants of Internet adoption, including the existing
telecommunication infrastructure, financial resources as well as human capital. The
characteristics of the human capital then can be seen as drivers of or barriers to Internet
Kiiski and Pohjola (2001) examine cross-country diffusion of the Internet using the
Gompertz model of technology diffusion. They include educational level as one of the
predictors in their study (Kiiski & Pohjola, 2001). Surprisingly, education does not show
statistical significance. This could be due to lack of variation across the countries, as they
examine only OECD members. Another possibility is that years of schooling is not the
best education variable to be used as a predictor of Internet adoption. Both of these
explanations could be true because when the regression analysis is replicated on a world
sample of countries and education is measured by university attendance, it becomes
College education is critical in the new communications era. According to the
World Bank, basic education is important overall, but "new, information-based
technologies are more demanding in skills for diffusing, interpreting, and applying
knowledge" (World Bank, 1999, 42). The report additionally notes that "countries at or
near the technological frontier need strong tertiary education and research institutions to
compete in the creation of new knowledge" (World Bank, 1999, 42).
English language proficiency is another important factor for Internet adoption since
the World Wide Web is still dominated by English-language Web sites (Global Reach,
2000), even though projections indicate that Chinese will become the dominant Web
language by 2007. If more users speak English within a country, they are more likely to
search for the predominantly English-language Web content. Sadowsky (1993), for
• Even though the domination of the English language is expected to wane, it was still by far the
predominant online language at the time this study was conducted.
example, argues that the ability to find online content is critical for Internet popularity in
developing countries. Interestingly, the rapid growth of Internet use in Bolivia has been
connected to the increased Spanish-language online content (Minges, 2001). English
language fluency facilitates not only content retrieval, but also computer and software
Wallraff (2000, 61) contends that "most people like new technology better when it
speaks their own language." Even though the English language has achieved a global
status, the proportion of English-speaking people is expected to shrink to less than five
percent in 2050. (Wallraff, 2000).
Caselli and Coleman (2001) include the fraction of the population who speak
English as a predictor of computer adoption. They test for the effects of English or
European language skills of the population. The language variable in this study is defined
as the proportion of those who speak English as a first language (Caselli & Coleman,
2001). The results of their study, however, show that fluency in English is not statistically
Hargittai (1999) defines "human capital" as related to Internet usage to include
educational level and English language proficiency. The addition of human capital
significantly improves the fit of the regression model. Hargittai (1999) finds that
education is positively correlated with level of Internet adoption (i.e., the higher the
educational level in the country, the higher its Internet connectivity). On the other hand,
lower education levels may prevent mass adoption of the Internet.
Most computer commands, instructions, and help files are in English.
Similarly to Hargittai (1999), Kiiski and Pohjola (2001) include English language
proficiency in their model. Hargittai' s study shows no significance for the English-
language variable. Even more unexpectedly, the English language variable in the Kiiski
and Pohjola (2001) analysis has a negative sign. It could be inferred that at this stage of
adoption among the OECD group of countries, English language is not a significant
The framework proposed here defines audience factors as including both
educational level and level of English language proficiency. If data permit, both will be
used as predictors for the level of Internet penetration in the post-communist countries.
These are the two most important audience characteristics identified in current literature
on new communication technologies adoption.
Finally, it has been argued that culture affects Internet adoption in a country (Elie,
1998; Maitland, 1998). Elie gives the example of European and Asian former Soviet
republics, claiming that GDP and telecommunications infrastructure cannot explain the
differences in their Internet positions. Maitland (1998) argues that culture should be
included as an explanatory variable of the adoption of interactive technological
innovations, such as the Internet. She argues that adding cultural differences makes the
understanding of Internet diffusion across countries more robust (Maitland, 1998).
Researchers acknowledge that the construct of culture is very complex (Jones,
1997; DiMaggio, 1997; Sondergaard, 1994; Tayeb, 1994). Writing in Britain in the late
1950s, Raymond Williams conceptualized one of the first definitions of culture. He
wrote: "Culture is ordinary: that is the first fact. Every human society has its own shape,
its own purposes, its own meanings. Every human society expresses these, in institutions,
and in arts and learning" (Gray & McGuigan, 1997, 6). Broadly defined, culture refers to
the values, religious beliefs, ethics, institutions, customs, and traditions shared by a group
of people. Thus, cultural traits are innate to all individuals living in a social system. They
are embedded and transparent within the culture.
DiMaggio (1997) among others points out that culture is a very complex concept.
Sociology and psychology researchers have looked at various dimensions of culture and
cognition. DiMaggio (1997) suggests that culture is fragmented across groups. The move
from understanding culture as a unified to fragmented phenomenon "makes studying
culture much more complicated" (DiMaggio, 1997, 265).
Hofstede (1980), however, sees culture as cohesive and manifested in national
societies where people within the country share certain beliefs and attitudes. This
conceptualization supports Williams' definition of culture (Gray & McGuigan, 1997).
Culture then exists at the societal level as an aggregate of individuals' shared beliefs and
Hofstede (1980) developed four dimensions to measure differences across
cultures: power distance, uncertainty avoidance, individualism and collectivism, and
masculinity and femininity. The cultural dimension which relates to the adoption of new
media and the Internet is uncertainty avoidance (Hofstede, 2001). Broadly defined,
uncertainty avoidance is "the extent to which the members of a culture feel threatened by
uncertain or unknown situations" (Hofstede, 2001, 161). It follows that countries with
higher uncertainty avoidance would be more resistant to the adoption of the Internet
compared with low uncertainty avoidance countries.
Katchanovski (2000) examines the influence of culture on economic growth in the
post-communist societies. He concludes that cultural differences do affect growth both
directly and indirectly. Katchanovski (2000) derives a cultural index on the basis of five
aspects of culture. His factor analysis identifies one factor consisting of civil society
index, religion, historical experience, and business index, which is labeled Western
Culture Index. The addition of cultural variables improves the fit of the regression model,
with the Western Culture variable showing the highest standardized regression
Domanski (2000) looks at religion and its effects on modernization in Eastern
Europe. The results of the study show a correlation between religiosity and social
stratification. It is clear that different religions are dominant in the different countries.
Based on 1993 data from six Eastern European societies, Domanski (2000) argues that
Poland is the most religious and the Czech republic the least religious of those. The level
of religiosity across all post-communist nations is expected to vary even more.
Cultural traits have been given as an explanation for differences in Internet
adoption among Western European countries (Forrester Research, 2000). Several studies
support the argument that culture influences the diffusion of technology in a country
(Maitland, 1998; Rey, 1998). Some studies divide Western European countries depending
on their geographic location (Beilock & Dimitrova, 2003; Kiiski & Pohjola, 2001). Kiiski
and Pohjola (2001) include a dummy variable for OECD countries depending on whether
they are located in southern or northern Europe (N=23). Beilock and Dimitrova (2003)
regress Internet usage rates against the percent of Roman or Orthodox Catholics in each
Western European country. Interestingly, these cultural variables show statistical
significance in both studies. In fact, the model fit improves once the regional dummies
are added in the Kiiski and Pohjola study (2001). Therefore, they conclude that there are
certain cultural factors that play an important role in the process of Internet diffusion.
Drawing from diffusion of innovations, Maitland offers five propositions how
culture can influence the adoption of interactive technologies at the country level.
Maitland positions gender equality as a social norm. Interestingly, she argues that the
diffusion of interactive networks, such as the Internet, will be higher in countries with
higher gender equality. Maitand (1998) extends the notion of "cosmopoliteness" from
diffusion of innovations to the country level. Basically, she suggests that openness of
society can be measured by the country's ethnocentrism, and that cultures low in
ethnocentrism will begin the diffusion of interactive networks earlier. This argument has
been supported by others as well (Rey, 1998). The way of measuring countries'
ethnocentrism can be problematic, however.
Religion is an important part of culture. In a study of economic reform in the post-
communist world, Fish (1998) includes religion as a determinant. First he uses three
categories for religious affiliation, but his ANOVA comparison of means indicates that
there is, in effect, no statistical difference between the Muslim/Buddhist and Eastern
Orthodox group. The analysis shows that the number of religion categories can be
reduced to two, namely the Catholic/Protestant countries as group 1 and the
Muslim/Buddhist and Eastern Orthodox countries as group 2 (Fish, 1998). Western
Christianity arguably shows similarity with Western culture/societies. The results in the
Fish (1998) study, however, show no effect of religion on economic growth.
The fifth dimension of our Internet diffusion model will be cultural factors. These
will be represented by dominant religious composition of the population. Incorporating
this variable with a focus on the three dominant religions also allows for measuring a
particular aspect of culture- Westernization-though the Western Christianity variable.
The diffusion of innovations literature reviewed above posits that diffusion is a
linear process. It identifies several individual characteristics that are critical for Internet
diffusion, namely income, educational level, cosmopoliteness and innovativeness traits.
The income and education levels of the audience are critical in the adoption process.
Diffusion of innovations research also shows that externalities are important for Internet
adoption in developing countries. Government policy and preexisting technology are
among the determinants identified in diffusion of innovations literature.
New media technologies research, too, supports the notion that the political climate
and technological infrastructure are critical for Internet use. Past research suggests that
economic factors remain one of the most significant determinants of Internet adoption at
the country level. New media technologies studies also show that audience characteristics
(human capital) as well as cultural factors play an important role in the diffusion process.
Thus, a multitude of factors emerge as significant for the Internet diffusion process
at the societal level. The five sets of factors identified in diffusion of innovations and new
media technologies literature constitute the conceptual framework used in this study.
1 . Economic factors
2. Political climate and policy factors
3. Technology/Infrastructure factors
4. Audience factors
5. Cultural factors
This five-dimensional analytical framework for the study of Internet adoption at the
country level is proposed and tested in this dissertation. Specific variables pertaining to
each set of factors are described in the following chapter.
Rogers posits that innovations have the following five attributes: relative advantage
(whether the innovation is better than previous ones), compatibility (whether it is
compatible with previous technologies), observability (how hard it is to use the
innovation), trialability (whether a person can try out the innovation), complexity (how
difficult it is to use it) and observability (whether the results of the innovation are readily
observable). He adds reinvention as the sixth additional attribute. Reinvention refers to
cases when an innovation is used in a new way/for completely new purposes.
These six innovation attributes and the five dimensions identified here interplay
with each other. The framework proposed above includes economic, political climate and
policy, technology/infrastructure, audience characteristics, and cultural factors. These are
conceptualized as country-level determinants of Internet adoption. Below, some possible
influences on Internet adoption patterns as a result of the interactions between the two
models are described.
The Internet is a very complex innovation. It can be used for a number of different
purposes, ranging from email to data searches to online marketing. The relative
advantage of the Internet will be higher for people who can see its value as a marketing
tool or even as a way to market their products abroad. In fact, countries in the lower
middle income range (such as many of the former socialist republics) may gain more by
using the Internet for import and export of various goods or services. In other words, the
relative advantage of the Internet will be higher for countries with lower national
incomes, which also tend to be less involved with international trade.
The Internet is not compatible with almost anything that existed prior to it. That
means countries with very low Internet may need to have special training programs for
potential users to show them how the Internet works. Countries with lower GNP levels
are less likely to spend money on training courses. The combination of lower income and
lower compatibility will make the adoption process in those countries for the Internet
slower, compared with innovations with higher compatibility in a country with the same
The benefits of Internet use are readily observable. Messages travel instantly and
information (usually) is downloaded very fast. Countries with lower income levels,
however, do not take advantage of higher speed Internet as much as they lack broadband
technology (ITU, 1999). Thus, the results of Internet browsing, emailing, downloading
and so on can be a little slower. If the user does not have a basis for comparison though,
this interaction is unlikely to have an effect on the rate of Internet adoption. The more
observable the results are to the users, the more likely they would be to go online again in
The trialability of the Internet as an innovation is also likely to affect its adoption.
If a person cannot try/does not have access to an Internet terminal, they are unlikely to
adopt it. Again, poorer nations have lower Internet penetration. Especially in rural/
remote areas, people have no access to the Internet. Countries with lower national income
will have less Internet connections/terminals available for public use, which means that
the potential adopter will have fewer chances to try out the Internet. This means they will
be less likely to adopt compared to persons in countries where the Internet is widely
People in more affluent societies tend to have access to more technologies and
other communication innovations (e.g., personal digital assistants or PDAs). In a way,
they are more likely to reinvent the Internet from say, just a typewriter or email box, to
use for online banking. (Online banking, of course, is related to a whole network of
technologies, software products, bank accounts security and other issues).
Some governments may adopt policies facilitating Internet penetration, especially if
they perceive the ability to receive public feedback online as valuable. Thus, the
relationship here is reversed: those political leaders who see the Internet as a tool for
encouraging public participation may be more likely to pass legislation that promotes
Internet adoption. Alternatively, those governments which see the Internet as a threat are
likely to adopt measures to limit its use.
As my dissertation proposes, privatization and liberalization policies can facilitate
Internet development and thus make it more available for people to try. This interaction
shows that for countries with more favorable market policies, the trialability of the
Internet will be higher. This interaction is likely to lead to faster adoption rates. Also, if
the beneficial results of Internet use are obvious for the policy makers, then adoption will
be promoted by favorable policies.
People who live in a country with limited communication technologies will see
more relative advantages of the Internet, as it is likely to be one of the few ways to
connect with people or get information. People who live in country with extensive
technological infrastructure will find the Internet more compatible, will have an easier
time using the Internet and trying it out in general, will find it less complex and may even
be more likely to reinvent its use because of their previous experience with various
Those who can read English will be able to find much more information online and
thus will perceive the Internet as more valuable. A person who can only read Macedonian
is limited in the amount of online content he/she can locate compared to another person
fluent in English. If the person uses the Internet for email only, that will not be a major
Arguably, people with higher education levels will see more value in Internet
information, particularly highly specialized information in their professional areas. In
addition, more educated people will be more likely to want to have professional contacts
with others around the world. Thus the Internet with its email and bulletin board
capabilities can be considered even more valuable.
A person with higher education is more likely to have used a personal computer
(PC). Therefore, such people will find the Internet more compatible with their previous
experiences as opposed to those who have never used a PC.
Some religions may view the Internet as more valuable than others. More closed
societies definitely are not as likely to embrace the Internet compared with more open
societies. Also, societies which have been isolated in terms of information from the
outside world may see the Internet as fulfilling an acute need—they will then perceive the
Internet as more valuable.
Some societies are more open to new ideas and find it easier to adopt new
technologies. Thus members of more open cultures may find it less complex to use the
Internet and also more compatible with their cultural predispositions. Certain cultural
traits can affect the way people use and come up with reinvention ideas for various
This chapter outlines the methods for testing the five-dimensional analytical
framework proposed in the previous chapter. The main research question addressed in
this study is what factors affect Internet adoption in the post-communist countries. Six
testable propositions are made. The study aims to identify the most significant predictors
of Internet adoption and explain a significant portion of the variation in Internet use.
Operational definitions, data collection methods, and hypotheses are proposed below.
The method chosen is governed by the primary research question and the nature of
the data. The research design is described next.
The literature review identified five sets of factors that affect Internet diffusion at
the country level. These are economic factors, political climate and policy,
technology/infrastructure, audience characteristics, and cultural factors. To use these
factors in a systematic, multivariate analysis, they need to be further defined as specific
variables pertaining to each set of factors. The operationalization of variables is shown in
Table 4-1. The study uses secondary data from a number of different sources. Data
sources and collection methods are provided below. The study employs t-test, multiple
regression, and Tobit analysis to test hypotheses and examine the significance of the set
of explanatory variables. Theoretical propositions and statistical procedures for testing
3le 4-1. Definition of variables ir
i the proposed model of Internet diffusion.
users per 10,000
Internet in a
The total domestic
and foreign value
added claimed by
residents within or
converted to the
U.S. dollar value of
the goods and
services which can
be purchased within
the country in the
Number of years
since the incumbent
has been privatized,
either fully or
(part of the
Level of civil
based on 14 criteria
telephone lines per
people plus the
number of mobile
phones per 1,000
Even though the World Development Indicators Report was issued in March 2000, most of the data are
from 1999 and 1998.
?le 4-1. Continued.
Tertiary percent of
relevant age group,
as a base: 1 if
otherwise; 1 if
The study uses secondary country-level data from a number of sources. It is
important to underscore that comparative data for the post-communist countries are
difficult to find. Fish notes that there "exists no reliable cross-national survey" that
encompasses all post-communist countries (1998, 37). Aggregate data from international
organizations—notably the International Telecommunication Union (ITU) and the World
Bank (WB)-are used in this study. Thus, definitions of indicators are drawn directly
from the original data source. Alternative data sources are also used to cross-check the
telecommunications data in particular.
The main data sources as shown in Table 4-1 are: Freedom in the World published
by the Freedom House; PrivatizationLink of the World Bank; World Development
Indicators published by the World Bank; and the World Telecommunication Indicators
Database published by the ITU. National government sources are also used in cases when
secondary data are not available from the international organizations noted above. For the
religion variable, the work of Fish (1998) is used as the main data source and cross
checked against the CIA World Factbook and country data from the U.S. State
Department Web site. Additional sources are noted in the statistical analysis.
It is critical to determine which potential variables best reflect the broad set of
factors identified in the literature review.
The main economic indicator to be used is per capita income. Gross national
product (GNP) per capita is the variable chosen in this model. This standardized variable
is a key indicator of national economic development and allows easy comparisons across
countries (Kennedy, 1998). The GNP measure includes the total domestic and foreign
value added claimed by residents as well as net receipts of primary income (both
compensation of employees and property income) from nonresidents (World Bank,
2000). GNP is calculated in U.S. dollars, using the Atlas method for conversion.
However, GNP in terms of purchasing power parity (PPP GNP) is more appropriate for
this study as it reflects the citizen's ability to buy goods and services (Arquette, 2002;
Beilock & Dimitrova, 2003), such as Internet connection or dial-up services. PPP GNP is
the U.S. dollar value of the goods and services, which can be purchased within the
country using personal income in the local currency. Thus, an "international dollar has
the same purchasing power over GNP as a U.S. dollar has in the United States" (World
Bank, 2000, 13). The income variable will be log-transformed, based on Rodriguez and
A few methodological notes are in order. First, the GNP data for Bosnia and
Herzegovina and for Ukraine come from the next edition of World Development
Indicators , using 1999 data-not 1998. Yugoslavia's GNP data is based on data for
Macedonia, because of the inherent similarity between the two transition economies.
Political climate and policy variables
The political climate is measured by the level of democratization in the post-
communist country, which is demonstrated by the level of civil liberties. Thus, a good
proxy variable for the level of democratization in the country is the Freedom House
ranking of the level of civil liberties. This ranking is based on 14 different criteria, which
include freedom of expression and belief, free and independent media and freedom of
cultural expression. The Freedom House has collected data on civil liberties within
countries since 1972. It publishes an annual assessment of the state of freedom within a
country by assigning a score to each state worldwide. The civil liberties ratings range
from 1 to 7. A rating of 1 refers to a country considered "Most Free" while a rating of 7
denotes "Least Free" countries. The scores will be inverted for analysis so that higher
rankings indicate higher levels of civil liberties in society. The inverted ratings also make
interpretation of the regression coefficients easier.
The civil liberties variable is a proxy for the level of democratization in the
country. In other words, the higher the level of civil liberties, the more democratic the
country is. Similarly to the Rodriguez and Wilson (2000) study, the civil liberties variable
is treated as interval even though it is an ordinal variable, similarly to the Rodriguez and
Wilson (2000) study. The difference between a country with a score of 5 and a country
with a score of 1 is not exactly 5 times, but this is assumed to be a good approximation
established by the Freedom House foundation.
The policy variable used is the length of telecommunications privatization. At first
glance, privatization can be seen as an economic variable. However, it belongs in the
political climate and policy category as it is a direct result of the decisions of the political
leaders in the country. The specific sector of interest in this study is telecommunications.
There is no consistent cross-country data on overall telecommunications policy in the
region. Thus, the study uses privatization of the incumbent telecommunications operator
as the best proxy for telecommunications policy in the country.
In this study, we assume that earlier privatization of the main telecom operator is an
indication of successful national policy in the area of telecommunications. In Bulgaria,
for example, the privatization of the main telecom operator, BTC (Bulgarian
Telecommunications Company), has been seen as one of the major privatization
transactions on a national level, comparable with that of the national electricity operator
and major tobacco companies. The fact that BTC's privatization has been extremely
difficult and slow is an indication of poor policy making on the part of the Bulgarian
government. This policy has allowed BTC to continue its monopoly. The reverse is true
in the case of Hungary: The major telecom operator MATAV was privatized in 1993 and
is considered a major sign of the success of Hungarian telecommunications
The definition of the telecommunications privatization variable is number of years
since the incumbent telecom operator has been privatized, either fully or partially, at the
end of 1999. This operationalization provides a method for determining the length of the
telecommunications privatization in the respective countries. The length of privatization
BTC's still pending privatization has not been completed as of end of 2002.
is clearly related to telecom deregulation in these countries, which makes it a good proxy
variable to be used in this research.
A number of variables that measure Information and Communication Technologies
(ICT) infrastructure exist. For the purposes of this study, only telephone infrastructure is
included, even though cable, TV, and other technologies can also play an important role
in Internet development. Broadband technology is not yet a viable option for Internet
connection in the post-communist countries. The telephone provides the main mode of
network connectivity in the region and has been used as a determinant of Internet
adoption in a number of studies (Beilock and Dimitrova, In press; CDT, 2000; Hargittai,
A logical possibility for an infrastructure variable in this model would have been
the number of computers per capita. As mentioned in Chapter 3, we cannot account for
Internet users if we do not have computers in the first place. However, lack of data for the
majority of the countries of interest prevents us from using that variable.
The most significant infrastructure variable regarding Internet usage is the number
of telephones in the country. Thus, the main infrastructure variable used here is the
number of residential phones per capita. This is measured by the International
Telecommunication Union (ITU) as telephone mainlines per 1 ,000 people. This measure
includes all telephone lines connecting a customer's equipment to the public switched
telephone network (PSTN).
Another potential measure of political climate and policy would have been the level of economic freedom.
Economic freedom is commonly measured by the Economic Freedom Index, which is based on 1 areas,
including government intervention, property rights, and foreign investment. However, economic freedom
does not directly reflect openness in the telecommunications sector, which is of particular interest in this
However, mobile phones play an important role in the technological advancement
of the post-communist societies. The growth of the mobile market in the post-communist
countries has been seen as a basic representation of the overall market demonopolization,
as noted in Chapter 3. Also, technology today allows users to connect to the Internet via a
mobile phone. The number of mobile phones per 1,000 population is then combined with
the number of residential phones per 1,000 population into one independent variable
called teledensity. The only disadvantage of using this composite variable is that we
cannot compare the separate effects of mainline telephones and wireless.
Two critical characteristics of the audience as Internet users were discussed in the
literature review. The most important variable in this category is educational level. Past
research shows that early adopters of the Internet tend to be more educated and usually
hold a college degree (Atkin et al., 1998; Howard et al., 2001; ITU, 1999). Thus, the
study uses educational level as measured by the gross enrollment ratio in tertiary
education of relevant age group. This variable shows how many people in the relevant
age bracket are attending college in a particular country.
As the literature review showed, people who speak English are more likely to go
online and have an easier time browsing the Web. The best proxy variable for English
language proficiency is looking at the number of students who take English as a foreign
language in school. One way to measure it is as the percentage of pupils in secondary
education learning English. However, such data are limited only to current or future
members (the so-called candidate countries) of the European Union. Thus, English as a
second language is not included in the final regression model, but is recommended for
Arguably, the best cultural measure is the dominant religion in the country. As Fish
observed about the post-communist region, "in the overwhelming majority of countries, a
single religious tradition clearly predominates" (1998, 41). National religion is a stable
variable, not quick to change (Fish, 1998). Using data from Fish (1998) and the CIA
Factbook (2000), the study incorporates religious affiliation as a cultural variable.
Orthodox Christianity is the most common religious group in the sample. The other
two major groups are Western Christians and Muslims. It is argued that these three
groups are substantially different from each other (Fish, 1998). The following nine post-
communist countries are predominantly Western Christian (Catholic or Protestant):
Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, and
Slovenia. The majority of the countries are predominantly Eastern Orthodox (Armenia,
Belarus, Bulgaria, Georgia, Macedonia (FYROM), Moldova, Romania, Russia, Ukraine,
and Yugoslavia (Serbia and Montenegro). Therefore, Eastern Orthodox religion will be
used as a base group for the dummy variables (coded as 0). The remaining nine countries
are predominantly Muslim, with the exception of Mongolia. 4 Thus, two dummy variables
will be employed to represent the three major religions in the region.
The dependent variable of interest in this analysis is Internet users per capita. This
indicator is measured by the International Telecommunication Union (ITU) as the
estimated number of Internet users based on the number of Internet hosts in the country.
The dependent variable in this study will be transformed as a natural logarithm of the
Mongolia is a predominantly Buddhist country, but is included in the Muslim group in this analysis, based
on Fish (1998).
number of Internet users because the distribution of that raw variable is skewed.
Logarithmic transformations are especially useful in cases when the normality of the
distribution is violated, usually resulting from the magnitude of the changes in the
observed variable. Therefore, the interpretation of the beta coefficients is in terms of
ratios, not in additive terms. When the data have been log-transformed, the slope
indicates a ratio of change (increase or decrease) in the dependent variable.
The number of Internet hosts is collected by Network Wizards on a biannual basis.
These data are used to estimate the number of Internet users per country. Internet hosts
and Internet users are the most commonly used measures for country-level Internet
penetration (Press et al., 1998). For complete explanation on data processing of the raw
Internet host data, check the World Telecommunication Indicators 2000/2001 edition or
visit the ITU Web site (http://www.itu.int/ITU-D/ict/publications/wti2000-01/). A brief
The number of Internet hosts per capita refers to individual computers connected to
the Internet—i.e., computers with an actual IP address. Network Wizards collects data on
active Internet hosts worldwide. This measure of Internet usage has been criticized in the
past (OECD, 1998a; Minges, 2000; Press, 1997; Zook, 2000). One reason is that Internet
hosts are misleading if more people use the Internet at Internet cafes rather than at a home
computer. In addition, Internet host figures are derived on the basis of country code top
level domains (TLDs) rather than actual physical location of the host. As Minges (2000)
explains, a host under the .RU country code domain can be located anywhere in the
world, not necessarily in Russia. By the same token, the so-called generic TLDs (.com,
A detailed description of Network Wizards' data collection methods can be found at
.edu, .gov, .int, .mil, .net, and .org) can be located in any country. 6 However, Internet
hosts are a conservative measure of Internet connectivity (Hargittai, 1999;
http://www.matrixnetsystems.com). Even though this measure has some shortcomings
(Minges, 2000; Press, 1997; Zook, 2000), it remains the most common measure of
Internet usage across countries (Press et al., 1998).
Another possible dependent variable is the number Internet subscribers per
telephone line. A study by Dasgupta, et al. (2001) used the number Internet subscribers
per telephone mainline as the dependent variable measure of national Internet use. Such
data are collected by the Economist Intelligence Unit (EIU)'s Pyramid Research.
However, this measure of Internet use can be misleading as many users in developing
countries log on the Internet from cyber cafes or from work. EIU's data also seem to have
lower reliability than the Network Wizards data.
As previous research indicates, there is no absolutely reliable estimate of the
number of users per host per country, and Internet metrics in general have been
somewhat inconsistent (Minges, 2000; Press, 1997; Rood, 1999; Zook, 2000). Measuring
the level of Internet adoption is a challenging task for many reasons. One is that there are
qualitative differences in terms of usage. For example, some users have access to
broadband connection or more advanced software than others whoa re limited by a 28K
or 56K modem. A person using a dial-up modem to connect to the Internet has a different
experience of the Net than a high-speed user.
Additional Internet penetration indicators have been developed to include the
following six different characteristics for country-level Internet adoption: pervasiveness,
Only seven generic top-level domains exited at the time this study was conducted. More generic TLDs
have been licensed by ICANN today.
geographic dispersion within the country, sectoral absorption, connectivity infrastructure,
organizational infrastructure, and sophistication of Internet use (Press et al., 1998). The
combination of these six characteristics could provide a fuller picture of the overall
Internet penetration than using simply one measure, such as Internet users per capita.
However, collecting data for each of the six variables requires experts to evaluate and
rank each country for each characteristic. Therefore, such data are unavailable for most
countries in the world. Another disadvantage of such measures lies in the subjective
nature of the ranking decision. At the present time, Internet users per capita represents the
most reliable measure of Internet penetration that can be used. It has the added advantage
that its values are easily comparable across countries.
Arguably, the best way to measure Internet users per country is by using panel
members from a national sample. This involves systematic sampling techniques and is
more precisely connected to an individual user. In other words, data is collected from
users' home computers, not from network traffic information. In the United States, there
are two companies that collect such data: Jupiter Media Metrix and Nielsen NetRatings.
However, no such research companies exist in the region examined here.
Even the methods used by such market research companies involve some
inconsistencies, which makes data comparisons difficult. Minges (2000) argues that
national surveys of Internet users often look at different demographic groups, for
example. Many research companies collect data on 16+ and 18+ age group categories.
Minges (2000) argues that looking at adult populations only disregards an important part
of current Internet users, precisely the younger population. In addition to age
breakdowns, geographic breakdowns and category breakdowns (users at work, at home,
and at school) vary widely across research companies.
Network Wizards' Internet host data has been commonly used by researchers.
Companies may adjust their Internet host data using a variety of methods. Matrixnet is
one company that produces adjusted measures based on Network Wizards' Internet hosts
counts. Their approach is to adjust the raw data from Network Wizards using statistical
estimations for localizing the host data. 7
Several other sources provide their own estimates on Internet users per county. Nua
is an Irish-based company that reports the number of Internet users per country. They
collect survey data from national surveys conducted by market research firms or
journalism reports, and aggregate these data. In some cases, Nua can decide to use one
source over the other, if the source seems more reliable. The problem is that there is no
consistency between the methodologies used in different countries. In addition, national
surveys are often conducted infrequently and inconsistently. The Nua Web site says that
they consider a user anyone who used the Internet at least once for the past three months,
or sometimes, for the past six months. That makes the Nua Internet users data unreliable
and comparisons based on such data very difficult.
It is very hard to estimate how many people are using the Internet within a country
(Daly, 1999; Press, 1997; Zook, 2000). One of the challenges for developing countries in
particular is that people often use email from Internet cafes or cyber cafes. These types of
use are more prevalent there than they are in the developed countries. The conclusion is
that ISP subscriptions may underestimate the numbers of Internet users as a result (Daly,
A detailed description of Matrixnet's methodology can be found at
1999). Another obstacle is the unwillingness of ISPs to disclose user data as they are
facing competing ISPs (Daly, 1999).
Another company that tracks down the number of Web sites on the Internet is
Netcraft. However, there is no direct relationship between the number of Web sites and
the number of Internet users per country (Press, 2000).
The variables described above can be divided into environmental variables (system
factors) and internal variables (human factors), as shown in Figure 4-1.
Figure 4-1. Graphic model.
The environmental variables (system factors) include the economic variable (GNP
per capita), the political climate and policy variables (democratization and
telecommunications privatization), and the infrastructure variable (teledensity). The
internal variables (human factors) include the audience variables (educational level and
English language proficiency) and the cultural variable (predominant religion). In
essence, the internal variables are "embedded" in the individual Internet user while the
external environment variables are conceptualized as factors outside the individual. Thus
Internet usage is determined by the audience's internal characteristics as well as system
factors existing externally.
Multiple regression is the main method employed in this study. The basic
assumptions of regression analysis are reviewed briefly below.
Multiple Regression Technique
Regression analysis is a type of a multivariate technique that measures relationships
among a set of variables, and is a popular method in social science research (Kleinbaum
et al., 1988). It allows researchers to evaluate the relationship between a single,
continuous dependent variable and one or more independent variables. As Kleinbaum et
al. (1988) note, there are a five basic assumptions in linear regression:
1. Y is a random variable, with a probability distribution with a finite mean and
variance for any fixed value of the independent variable Xj
In this study, the observations included in the analysis constitute the population of
interest. Thus, the dependent variable-Internet users per 10,000 population—is not a
random variable per se. However, it can be argued that it is a random sample where each
country's probability of being picked equals 1. Therefore, we can treat the dependent
variable as having a probability distribution with a finite mean and variance for the values
of the independent variables.
2. Independence: Y follows a normal distribution, and the Y-values are statistically
independent from one another
It is logical to expect that in this model the dependent variable will follow a normal
distribution after a logarithmic transformation because the gaps between the Ys increase
as Y increases. It is also expected that the number of Internet users per capita in country
A will not affect the Internet users per capita in country B, i.e., the values of the
dependent variable will be statistically independent. Both of these assumptions will be
3. Linearity: Y is a linear function of X (i.e., the mean value of Y is a straight line in
regard to X)
This assumption states that the relationship between the dependent variable and the
independent variables is linear. The assumption is likely to hold true in this model based
on previous research and also on a brief look at the data. As GNP increases, the number
of Internet users also increases. Again, the validity of the assumption will be tested both
visually and statistically in the next chapter.
4. Homoscedascisity: The variance of Y is the same for any value of X
Another important assumption in regression analysis is that the dependent variable
will have the same variance for each value of X. Violations of homoscedascisity— i.e.,
heteroscedascisity, often occur in conjunction with violations in the normal distribution.
Therefore, it is expected that the dependent variable will exhibit equal variances if it is
5. Distribution: For every fixed value of Xj, Yj is normally distributed
The final assumption is that the distribution of the Ys will be normal respective to
the X values. There is no reason to expect violations in the distribution of the residuals.
In regression, the so-called least squares method "determines the best-fitting
straight line as that line which minimizes the sum of squares" of the residuals (Kleinbaum
et al., 1988, 49). Thus, regression analysis allows for estimation of both the direction and
strength of the relationship between variables.
The model proposed here is that the dependent variable Y (number of Internet
users) is a function of ECON, POLI, TECH, AUD, and CUL factors:
Y= / [ECON, POLI, TECH, AUD, and CUL]
It is expected that the regression model follows the basic regression formula
Y=a + P1X1 + p 2 X 2 + . . . + PiXj , where a is the intercept and the (3 is the slope of the
regression line. The Xs represent the independent variables/predictors. Clearly, when all
the Xs equal zero, Y=a. When X| increases by 1 (controlling for the other predictors), the
mean of Y increases by pV When changes in the values of Xj have no effect on the mean
of Y, then p\=0. It is inferred then that there is no relationship between the dependent and
Multiple regression is the best method to answer the question: What are the most
significant predictors of the variable Y? One of the advantages of multiple regression is
that it allows researchers to detect the relationship between X and Y while controlling for
a subset of other predictors. In other words, the partial contribution of the variable X can
be measured when other covariates are included in the regression model. Regression
generally looks for a linear association between the variables. It tries to fit the Y values
on a regression line by minimizing the (squared) distances between the observed value Yj
and the predicted value of Y.
The goal in regression analysis is to explain as much of the variation of Y as
possible. It is tested whether the regression function—using the combination of the Xs—
will explain more of the variation in Y than only using the mean of Y. The amount of
variation explained is expressed in an R-square value. The R-square value ranges
between and 1 . The closer the R-square value is to 1 , the more of the total variation in
Y is explained by the regression model. Since this analytical model is more
comprehensive than previous research, it is expected that the R-square will be high. No
direct comparisons of the R-square value will be possible, however, because no study to
date has examined Internet use in the same set of countries.
Several automatic selection procedures exist that help reduce the number of
explanatory variables to be used in a regression model. These include backward and
forward selection methods. Such stepwise regression methods allow researchers to select
a subset of the initial set of explanatory variables. That technique is especially useful for
exploratory research (Stevens, 1992). Forward stepwise regression begins by entering the
predictor that has the strongest partial contribution to the dependent variable Y. At each
subsequent step, another predictor is entered. The backward stepwise method is the
opposite: all explanatory variables are initially entered in the model. Variables that lose
statistical significance after entering the regression model are removed (in contrast to
forward selection where predictors, once included, cannot be removed from the model).
Backward regression was chosen as a stepwise method. First, all explanatory
variables will be entered into the model. Next, the backward selection procedure will
remove the variable with least partial contribution to the dependent variable Y. Therefore,
any of the predictors may be removed as insignificant from the final model at subsequent
steps. The criterion for removal will be F-test probability larger than .10. Also, the
backward regression procedure is set up so that variables removed earlier do not reenter
the model at subsequent steps. In this study, we begin by including all predictors in the
regression model. This complete model is used to test the proposed hypotheses. Next,
variables will be removed one by one, using backward regression. Thus, a final model in
which all explanatory variables are statistically significant will be derived. Comparisons
between the complete model and the reduced/ final model are offered based on R-square
and Adjusted R-square values.
Thus, backward regression is used to select the most important explanatory
variables for the final regression model. This statistical procedure, even though hardly a
panacea, also reduces the problem of multicollinearity as variables that do not bring
individual contributions (in other words, do not explain additional variation of Y) are
removed. However, it is important to note that stepwise regression has some inherent
issues as a regression type. The use of stepwise regression here is only justified by the
fact that this is an exploratory study.
Based on the literature review, the following hypotheses are formulated and
empirically tested, as reported in Chapter 5.
Internet adoption in the post-communist countries will increase over time. Based
on diffusion trends in other regions of the world-and particularly in Western Europe-we
expect steady growth in the number of Internet users.
Hypothesis 1 : The number of Internet users per capita in the post-communist
countries increased from 1995 to 1999.
Test: Paired-samples T-test for comparison of means: Yi99 5 < Y1999
Economic factors play a critical role as a driving force of Internet adoption. A
higher level of economic development will result in higher Internet penetration within the
post-communist countries. The key indicator of economic development is GNP per
Hypothesis 2: The higher the level of PPP GNP per capita, the higher the number
of Internet users per capita.
Test: H A : p\>0
Restrictive telecommunications policies and lower levels of democratization will
result in lower Internet diffusion. Conversely, more political freedom and faster
liberalization in the telecommunications sector will encourage further Internet
Hypothesis 3a: The higher the level of civil liberties, the higher the number of
Internet users per capita. (The civil liberties score will be reversed to show a positive
Test: H A : (3 2 >0
Hypothesis 3b: The longer the period of telecommunications privatization, the
higher the number of Internet users per capita.
Test: H A : (3 3 >0
More developed technology infrastructure will lead to greater levels of Internet
usage within the post-communist countries. Teledensity is the most critical infrastructure
determinant of Internet use. Thus,
Hypothesis 4: The higher the teledensity in the country, the higher the number of
Internet users per capita.
Test: H A : p* 4 >0
Audience characteristics, including education level and English language
proficiency, affect the levels of Internet usage in the post-communist countries.
Hypothesis 5: The higher the tertiary education ratio, the higher the number of
Internet users per capita.
Test: H A : p>0
Cultural predispositions affect the level of Internet use within the post-communist
countries. Dominant religion, as a major cultural factor, will affect Internet penetration.
Hypothesis 6: Differences in national religion will affect the number of Internet
users per capita.
Test: H A : p\#0; pVO
Multivariate quantitative studies often face the issue of high correlation between
the explanatory variables. In such cases, reduction of variables is probably the best
solution (Agresti & Finley, 1997; Stevens, 1992). Variable reduction is especially
desirable in this study, where a large number of predictors and a small number of
observations are examined.
The population of this study includes the countries of Eastern Europe, the former
Soviet Union, and Mongolia. The unit of analysis is the country. Statistical analysis will
allow us to make inferences about the relationships between the dependent variable and a
Due to lack of language data, the effects of English language proficiency on Internet adoption cannot be
number of independent variables. Parameters will be estimated and hypotheses tested in
the following chapter.
The number of observations in this statistical analysis is relatively small (N=28).
The specific set of countries includes: Albania, Armenia, Azerbaijan, Belarus, Bosnia and
Herzegovina, Bulgaria, Croatia, Czech Republic, Estonia, Georgia, Hungary, Kazakhstan,
Kyrgyzstan, Latvia, Lithuania, Macedonia (FYROM), Moldova, Mongolia, Poland,
Romania, the Russian Federation, Slovak Republic, Slovenia, Tajikistan, Turkmenistan,
Ukraine, Uzbekistan, and Yugoslavia (Serbia and Montenegro).
The relatively small number of observations is justifiable considering (1) the
exploratory nature of this study, (2) the strong theoretical base, (3) the
comprehensiveness of the model, and (4) the fact that it includes the population of
interest. The independent variables chosen have been identified as significant Internet
determinants in prior studies of cross-country Internet diffusion. There is very little
research on Internet adoption in the post-communist countries. Finally, the issues of
sampling do not apply here as the total number of countries— i.e., the entire population of
interest— is included in the study. Therefore, the results of this study cannot be criticized
because the observations include the entire population of interest, namely all the post-
The dependent variable in this analysis has an interesting feature: It cannot take on
negative values. Clearly, the lower end of the variable— number of Internet users— is zero.
Such variables are called truncated. Some econometric literature suggests that using
classical linear regression for such dependent variables may produce biased results
(Gujarati, 1995; Maddala, 1984; Tobin, 1958). In simple terms, the bias comes from the
fact that the assumption of equal variance of the dependent variable across all values of
the independent variables is violated when the values of the dependent variable approach
zero (Maddala, 1984). In such cases an alternative procedure called Tobit is
recommended. Using linear regression with truncated dependent variables is not a
problem when only few of the values are close to zero, however. In this study, the lowest
number of Internet users per 10,000 population is 3.13 for Uzbekistan. Only three
countries (Uzbekistan, Tajikistan, and Turkmenistan) have less than 5 Internet users per
In a study of worldwide Internet diffusion, Beilock and Dimitrova (2003) found
that the results of the two methods-Tobit and multiple regression-were, in effect,
identical. Tobit himself suggests that the results of the regression analysis are a close
approximation of the results produced by Tobit analysis (Tobin, 1958). In this particular
study, there is enough variability across the values of Y to assume that multiple
regression will produce unbiased estimates. The statistical tests for homoscedasticity
show no serious violation. Still, a Tobit analysis will be conducted as an additional step
to check the coefficient estimates of the regression against the Tobit estimates.
Another limitation in this study is the problem of multicollinearity.
Multicollinearity is evident across the bivariate correlations between the explanatory
variables (See Table 4-2). As expected, multicollinearity exists in several cases: more
affluent societies tend to have more developed telecommunications infrastructure (r=.84);
teledensity is also highly correlated with the level of democratization (r=72) and length
privatization (r=55). However, it will not be appropriate to eliminate variables identified
in the literature as significant only because of potential multicollinearity issues.
The multicollinearity problem is strongest between GNP per capita and teledensity.
The implication of such multicollinearity is that these two variables, if entered together in
the model, will affect each other's significance. However, the overall significance of the
F-test is uncompromized. Further, the stepwise regression method will partially take care
of the multicollinearity problem because variables that do not independently add to the R-
square will be removed from the regression equation.
Table 4-2. Correlation matrix for the continuous indep endent variables.
This study measures Internet adoption at a single point in time, except for
Proposition 1. The five-dimensional framework proposed here is tested by Hypotheses 1
through 6 examines the relationship between Internet use and a set of explanatory
variables at one point in time. Thus, this study is cross-sectional, which limits the
applicability of the findings for future stages of Internet diffusion.
Another methodological concern is the use of proxies. Data unavailability makes it
necessary to include proxy variables for a number of factors identified in the literature
review. In particular, it is very difficult to measure culture empirically. Admittedly,
religion is only one characteristic of cultural differences.
The data on Internet users per capita--my dependent variable —need to be
considered an approximation, an estimate of the actual number of Internet users in the
country. However, these are the best data on Internet use available at the time of this
research. Again, these numbers are best estimates and are more useful for comparison
purposes rather than in absolute terms. The concluding chapter offers more discussion on
the limitations of this study.
Finally, it is important to note the following changes in terminology took place
while this study was being conducted: The country of Yugoslavia has a new official
name. The new name adopted in March 2002 is Serbia and Montenegro. Also, GNP is
now called GNI (Gross National Income), according to the World Bank. This new label
was adopted in 2001. For the purposes of this manuscript, however, the two original
names are used.
Next, Chapter 5 describes the results of the statistical analysis and the hypothesis
The main goal of this study was to identify the most significant predictors of
Internet adoption in the post-communist countries. A t-test, multiple regression, and Tobit
analysis were performed to test the hypotheses proposed in the previous chapter. The
results showed that by applying the five-dimensional framework, a significant portion of
the variation in Internet use was explained. Before looking at the hypothesis testing,
descriptive statistics of this study are presented next.
The dependent variable used in this study was Internet users per 10,000 population
(IUR). This indicator is reported by the International Telecommunication Union (ITU) as
the estimated number of Internet users in the country, per 10,000 population. In this
study, the variable is log-transformed, using natural logarithm as a base, because the
distribution of the raw variable is skewed. This logarithmic transformation was
performed to alleviate the violation of the normality regression assumption. Variations in
the dependent variable are described below. It is important to note that when data have
been log-transformed, the regression slope indicates a ratio of change in the dependent
variable. Therefore, the regression coefficients below reflect a percent change in IUR.
The range for the transformed dependent variable lnRJR (Log of Internet users per
10,000) shows a nice dispersion. For easier interpretation, the raw (non-logged) numbers
are discussed here. Table 5-1 shows that the country with the highest number of Internet
users per 10,000 people is Estonia (Ranked #1). It has 1,383.5 users per 10,000
population, followed by the Slovak Republic (#2) with 1,300.7 users, and Slovenia (#3)
with 1,257.0. These countries are the Internet leaders among the former Soviet bloc
countries. The Czech Republic (#4), Hungary (#5), and Poland (#6) rank next. At the
other end of the spectrum, data show Uzbekistan with 3.1 users per 10,000 population (as
#28 at the bottom), preceded by Tajikistan (#27) with 3.3, and Turkmenistan (#26) with
4.6. These countries are the laggards in Internet adoption in the post-communist world.
The other three countries with fewer than 10 users per 10,000 population are Albania
(#25), Bosnia and Herzegovina (#24), and Belarus (#23).
There is clearly a wide disparity between the leaders and the laggards in Internet
adoption among the post-communist countries. The difference between Estonia at the top
of the ranking and Uzbekistan at the bottom is more than 400 times. Even a country in the
middle of the ranking such as Romania (#10) has 70 times more Internet users per 10,000
population than Uzbekistan (#28). Two republics of the former Soviet Union show
similar discrepancies: Russia (#12) has 19 times the Internet use of Belarus (#23). The
degree of difference between Kazakhstan and Belarus is smaller but still substantial:
Kazakhstan has over 4 times the number Internet users of Belarus.
The mean for the non-logged values of Internet users per 10,000 population is
290.85 and the median— the middle value-is 80.16, which indicates right-skewed
distribution. The percentiles show that 25 percent of the data points are under 10.66 and
25 percent are above 442.65 Internet users per 10,000 people, with Inter Quartile Range
of 433.99. The logged values range from .50 to 3.14, with a mean of 1.88 and standard
deviation of .85, as shown in Table 5-2.
Internet users per 10,000 people in 1999.
Bosnia and Herzegovina
Gross National Product
Gross National Product (GNP) per capita was used as a predictor variable.
Differences between GNP and GDP are typically relatively small and the two indicators
are often used interchangeably. GNP is defined by the World Banks' World Development
Indicators as the total amount of domestic and foreign value added claimed by residents.
The specific variable used here is GNP in terms of purchasing power parity (PPP GNP),
the U.S. dollar value of the goods and services that can be purchased within the country
using individual income in the local currency. In the post-communist countries, GNP
varies from a low of $1,041 for Tajikistan to a high of $14,400 for Slovenia. Again, the
relative differences show a considerable gap between top and bottom country in the
region. Because of its skewed distribution, the GNP is log transformed, using natural
logarithm as a base. The logged values for this variable-lnGNP-range from 6.95 to 9.57
(See Table 5-2).
Descriptive statistics of variables.
Eastern Orthodox (EST)
Western Christian (WST)
a. 11 'K has bee
n natural- logged.
See Chapter 4 (Methods) for complete description of variables and data sources.
This is a GNP PPP per capita measure and it has been natural -logged.
Due to lack of data, per capita figures for Yugoslavia ( Serbia and Montenegro) are based on Macedonia figures.
Teledensity combines both residential and mobile phones,
f. Mongolia was coded as Muslim due to its cultural similarity with this group.
A proxy for the level of democratization in the country is the Freedom House
ranking of civil liberties. The civil liberties scores range from 1 to 7. The variable used in
this study, DEM, was based on a reversed ranking of the Freedom Forum score. In this
study, a value of 1 means "Not Free" and a value of 7 reflects a "Free" country.
Variations across this predictor variable range from 1 to 6. In other words, none of the 28
countries in the model is ranked as completely free, according to the Freedom House
Foundation data. However, nine countries received a ranking of 6, Almost Free. Only
one country-Turkmenistan-received a ranking of 1, Not Free. Note that civil liberties is,
in essence, an ordinal variable, but it is treated as interval in this analysis as noted in the
Telecommunications privatization was measured as the number of years since the
incumbent telecommunications operator has been privatized, either fully or partially, at
the end of 1999. There is relatively little variation across this variable. Several of the
countries of interest have been rather slow in privatizing their major telecom operator.
Specifically, the telecom privatization variable equals zero in 17 (or 61 percent) of the
cases. The country that privatized its operator first was Estonia, followed by Hungary and
Latvia. The majority of countries, however, have not yet privatized or did so very
recently. This lack of variation may reduce the potential significance of the variable in
the regression testing.
The infrastructure variable was constructed by adding the number of mobile phones
per 1,000 population to the number of residential phones per 1,000 population. Not
surprisingly, data vary widely across the population of the study. As shown in Table 5-2,
the mean teledensity for the 28 countries is 226.79 phones (mobile and residential) per
1,000 population. The number for Albania is 32 (lowest teledensity in the region) in
contrast to Estonia where telephone penetration was 513 (highest teledensity in the
Education was measured by the tertiary education ratio, using data from the World
Development Indicators database. This variable shows how many people from the
relevant age group attended college. There is relatively little variation across this
variable: the mean for all countries was 27.72, with standard deviation of 10.07 (See
Table 5-2). The top country with a ratio of 45 is Estonia whereas Albania is at the bottom
of the list with a ratio of 1 1 .
Predominant religion was used as a cultural determinant. The frequencies show that
nine of the post-communist countries are predominantly Western Christian (Catholic or
Protestant), ten countries are predominantly Eastern Orthodox Christian, and nine are
predominantly Muslim/Buddhist. Mongolia is a predominantly Buddhist country, but it
was coded as Muslim, based on Fish (1998), due to its cultural similarity to this group.
To test the proposed hypothesis, two dummy variables were constructed. Western
Christian (Catholic or Protestant) was coded as 1 and otherwise in the first dummy
variable. In the second dummy variable, a country was coded as 1 if predominantly
Muslim or Buddhist and otherwise. Thus, Eastern Orthodox Christianity was the
category for religion.
Next, bivariate correlations between the dependent variable-log of Internet users
per 10,000 population-and the set of independent variable, excluding the religion
dummies, were run. The Pearson correlations are shown in Table 5-3. The variable with
the highest bivariate correlation coefficient is teledensity, followed by democratization.
Both of these variables are highly correlated with InlUR and both coefficients are well
above .80 whereas the correlation with InGNP is close to .80. The bivariate correlation
coefficients are lower for telecommunications privatization (r=.50) and education (r=25).
All coefficients are positive, indicating a positive relationship between IUR and the
predictors, without controlling for the other variables.
Table 5-3. Pearson correlations between dependent and independent variables.
After initial examination of the bivariate correlations, a multivariate regression was
conducted. The results of the multiple regression analysis are presented in the next
The classical assumptions of linear regression are: first, a linear relationship
between the dependent variable and the determinants; second, homoscedasticity (or equal
variance of Y for all Xs); and third, normal distribution of the dependent variable across
the values of the independent variables. All of these assumptions are satisfied in this
analysis, either initially or after subsequent transformation of variables. An examination
of the histograms as well as statistical tests showed no significant deviations.
jay pub Bjusog
The five-dimensional theoretical framework proposed the following function: Y= /
[ECON, POLI, TECH, AUD, and CUL], where the dependent variable Y is the number
of Internet users per capita. Y was defined as a function of economic, political climate
and policy, technology, audience, and cultural factors. The specific hypotheses stated on
the basis of this function are tested below.
Hypothesis 1 predicted that the average number of Internet users in the region
increased significantly from 1995 to 1999. A paired-samples t-test for comparison of
means was performed. Due to lack of data for 1995, only 19 countries are included. The
results of the t-test comparing Internet use in 1995 and 1999 are significant (t ( i8> =3.07;
p=.007). The t-statistic is positive, which indicates that the number of Internet users
increased significantly from 1995 to 1999. Thus, hypothesis one is supported.
Despite the general increase in the number of Internet users in 1999, variations
across the post-communist countries remained stark. Figure 5-1 shows the unequal
distribution of Internet users per capita based on 1999 data. It is obvious that the degree
of difference in Internet use between the countries in the region is substantial, with three
countries clearly at the top: Estonia, the Slovak Republic, and Slovenia.
The second hypothesis predicted that higher per capita income would contribute to
higher number of Internet users per capita. Table 5-4, Model 1 shows the results of the
multiple regression analysis. The results indicate that this hypothesis is supported. The
standardized beta coefficient for GNP (B=.23, p=.031) is positive and statistically
significant at the .05 level (See Table 5-4, Model 1). As expected, higher InGNP leads to
The third hypothesis predicted a positive relationship between the political and
policy variables and Internet adoption levels. Hypothesis 3a stated that the higher the
level of civil liberties in the country, the higher the number of Internet users per capita.
The results of the regression analysis show support for this relationship. As Table 5-4,
Model 1 illustrates, the standardized beta coefficient for DEM is positive and statistically
significant (B=36, p=.001). Thus, the regression results show strong support for this
Table 5-4. Regression results for Internet users. 3 ' b ' c '
~a. Dependent variable: Log ot Inlemet users.
b. The table reports Standardized Beta Coefficients with significance in brackets.
c. The coefficients are based on one-tailed tests.
Hypothesis 3b proposed that the longer the period of telecommunications
privatization, the higher the number of Internet users per capita. Even though the
coefficient for PRIV is positive, this variable is not statistically significant (B=.09,
p=.141). Thus, we cannot reject the Null hypothesis that p ? is larger than 0. Hypothesis
3b is not supported, as shown in Table 5-4, Model 1 .
Hypothesis 4 predicted that higher teledensity in the country will lead to higher
Internet use per capita. Even though the sign of the beta coefficient is positive (B=.24,
p=.148), the partial t-test in the regression table is not significant at the 95 percent
confidence level (Table 5-4, Model 1). Thus, the results of the regression analysis show
no support for Hypothesis 4.
The results of the regression analysis do not support Hypothesis 5 either. Higher
tertiary education ratio seems to be related to lower number of Internet users per capita
(B=-.10, p=. 1 15). However, the results for this predictor are not significant (See Table 5-
4, Model 1). Thus, we cannot reject the Null hypothesis and conclude that educational
level is not positively related to Internet use.
The last hypothesis predicted that differences in the predominant national religion
would affect the number of Internet users per capita. The results of the regression
analysis show partial support for Hypothesis 6, as reported in Table 5-4, Model 1. The
standardized beta coefficient for the Muslim dummy variable is negative and statistically
significant (B=-.26, p=.016) whereas the coefficient for the Western Christian group is
not significant (B=.01, p=.47). Thus we conclude that Muslim religion is significantly
and inversely related to Internet use, but Western Christianity is not, when examining the
post-communist countries. Bear in mind that the values of the religion dummies are
relative to Eastern Orthodox as the base and do not represent absolute values.
The most surprising result is that teledensity is not statistically significant in the
complete regression model. This could be partially due to the high correlation between
TEL and GNP. Another possible explanation for the lack of significance of TEL is that
the countries in the region are relatively close in terms of existing infrastructure,
compared with countries in sub-Saharan Africa and Western Europe, for example.
Teledensity may not be a constraint for Internet use in the post-communist countries.
Further analysis was performed to determine which explanation is more plausible. Only
income, democratization, and Muslim religion are statistically significant in the full
Table 5-5. Summary of hypothesis testing.
The percent of Internet users in the post-
communist countries increased from 1995 to
The higher the level of PPP GNP per capita, the
higher the number of Internet users per capita.
The higher the level of civil liberties, the higher
the number of Internet users per capita.
The longer the period of telecommunications
privatization, the higher the number of Internet
users per capita.
The higher the teledensity in the country, the
higher the number of Internet users per capita.
The higher the tertiary education ratio, the
higher the number of Internet users per capita.
Differences in national religion will affect the
number of Internet users per capita.
The results of the multiple regression analysis show that the six predictors
combined explain 92 percent of the variation in Internet use (See Table 5-4, Model 1).
The adjusted R-square is .896, which is also substantial. Three of the predictors in the
complete model-GNP, DEM, and MSL—are significant at the .05 probability level.
However, the other four predictors— PRIV, TEL, EDU, and WST-- are not statistically
significant. Therefore, subsequent regression models are developed to determine the most
robust model for these data.
Backward regression was employed to determine the best regression model, the one
employing the fewest predictors and having the strongest explanatory power. Adjusted R-
square values were used as a criterion, with higher Adjusted R-square indicating a better
model. At each step, the backward regression procedure removed the variable with the
least contribution from the complete model. The criterion used to delete variables was F-
test probability higher than .10. Using this procedure, the first variable to leave the model
was Western Christian religion (See Table 5-4, Model 2). At the next step the
telecommunications privatization variable was removed (Table 5-4, Model 3). Next,
tertiary education ratio was removed (Table 5-4, Model 4), with income, democratization,
teledensity, and Muslim religion remaining in the model. Of the seven initial variables,
only four were left in the reduced final model.
The four predictors that remained in the final regression model were income
(InGNP), democratization (DEM), teledensity (TEL), and Muslim religion (MSL). These
four predictors, collectively, explain 91 percent of the variation in Internet use. The
adjusted R-square for the final model is .897, which is slightly higher than the value in
the complete model. It is also evident that the F-test from Model 1 (complete model) to
Model 4 (final model) increased. The ANOVA tables for the each of the four regression
models are presented below with a comparative discussion.
The complete model of the multiple regression is shown in Table 5-6. Based on this
model, the following regression equation line is derived:
InlUR = -1.458 + .3021nGNP + .195DEM + . 027PRIV + .001TEL - .008EDU +
.020WST - .462MSL + e
Table 5-6. ANOVA Table foi
Complete Model.*' b,c
a. Dependent Variable: Log Internet users
d. Because education was not showing a significant effect, and we suspected some indirect relationship
with GNP for that, the GNP variable was removed from the model and all coefficients were
recalculated. However, neither the size nor the direction of the coefficient for education changed.
e. Instead of tertiary education ratio, high school education was used as an alternative education measure.
However, the coefficient remained negative and insignificant.
f. Because Western Christian religion (WST) was not showing a significant effect, and we suspected
some indirect relationship with Western Christianity and teledensity. Therefore, the teledensity
variable was removed from the model and all coefficients were recalculated. However, the WST
variable remained insignificant.
The equation for the complete regression model shows that one percent increase in
national GNP leads to .30 percent increase in IUR since both the dependent variable IUR
and the income variable GNP have been logged. An increase by 1 in the civil liberties
score results in a .195 percent increase in Internet users per 10,000 population. If
privatization is a year longer, that leads to a .027 percent increased in IUR given that the
other variables are held constant. Adding one unit to teledensity increases the mean of
IUR slightly by .001 percent. Increasing education by one unit leads to a .008 percent
decrease in the mean of IUR. Being predominantly Western Christian leads to .02 percent
increase in IUR. On other hand, being a predominantly Muslim society leads to a .462
percent decrease in the number of Internet users per 10,000 population.
Table 5-7. ANOVA Table for Model 2. ab,c
a. Dependent Variable: Log Internet users
The second model includes only six variables since the first step of the backward
selection procedure removed the Western Christianity variable. There are only minor
changes in the regression coefficients from Model 1 to Model 2. The only change worth
noting is the increased significance of MSL after the removal of WST. The overall R-
square for Model 2 remains .92 while the adjusted R-square increases from .896 to .901
(See Table 5-4, Model 2). This was expected because of the reduced number of variables
in Model 2.
At the next step of the backward selection procedure, the telecommunications
privatization variable was removed. The ANOVA table for Model 3 (Table 5-8) shows
that the two strongest determinants seem to be democratization and teledensity, as
indicated by their standardized beta coefficients. The R-square of the regression model
goes down slightly (from .923 in Model 2 to .918 in Model 3) as does the adjusted R-
square (from .901 in Model 2 to .899 in Model 3). Still, more than 90 percent of the
variation in Internet users is explained by the present model. The adjusted R-square also
Table 5-8. ANOVA Table for Model 3. a ' b ' e
a. Dependent Variable: Log Internet users
The final model is derived through removing the tertiary education variable using
the backward selection procedure. Table 5-9 shows the ANOVA table for the final
model. Model 4. Western Christian religion, telecom privatization, and education have
been removed from the complete regression model. Thus, the four determinants left in the
reduced model are income, democratization, teledensity, and Muslim religion. The
standardized beta coefficients indicate that democratization is most influential among
those, followed by teledensity, national income, and Muslim religion. This final model,
Model 4, has an R-square of .912 and an Adjusted R-square value of .897. The final
model also has the most significant F-test (F=59.77, p=.000) as shown in Table 5-4,
Model 4. The regression line derived from Model 4 is the following:
InrUR = -1.683 + .2841nGNP + .222DEM + .002TEL - .347MSL + e
Interpreting the partial effects of the four remaining variables, it is clear that one
percent increase in national GNP leads to a .284 percent increase in IUR. An increase by
1 in the civil liberties score results in a .222 percent increase in Internet users per 10,000
population. Similarly, an increase by 1 in teledensity increases the mean of IUR by .002
percent. Finally, being a predominantly Muslim country leads to .347 percent decrease in
the number of Internet users.
Table 5-9. ANOVA Table for Model 4. a ' K c
a. Dependent Variable: Log Internet users
Which of the four models presented above should be selected as the best model?
The second model, which included all predictors except the Western Christianity variable
removed at the first step of the backward selection procedure, had the highest adjusted R-
square. If we only look at adjusted R-square value then, Model 2 should be considered
superior to Model 4. The final model-Model 4, however, is more parsimonious, as it
includes fewer predictors and explains almost as much of the variation in IUR as Model
2. In fact, the F-tests for Model 4 is highest compared to the previous regression models.
Based on the F-test statistic, we conclude that the final model—Model 4-could be seen as
better than the previous regression models.
The results of the final regression model, again, were derived by a backward
selection procedure so once a variable was removed, it did not come back into the model.
Thus, multicollinearity among variables may have affected the selection order of the
remaining variables. The backward regression, however, helped select the most
significant determinants of Internet use in the post-communist countries. Due to
multicollinearity, the individual beta coefficients in the multiple regression analysis
should be interpreted with caution. However, multicollinearity does not jeopardize the
explanatory power of the overall model. Limitations of the study are addressed in the
Among the seven independent variables, income and teledensity are the two
variables most highly correlated. Putting them together in one model raises the question
whether there is some indirect relationship between the two variables. However, previous
literature shows that both factors are important, which makes it necessary to include both
variables in the model. A possible indirect relationship between income and teledensity
can be expected. However, a study found that adding an interaction term for these two
variables did not matter (Dimitrova, In press).
The dependent variable in this study— number of Internet users— cannot take on
negative values. Therefore, IUR is a truncated variable and invites the use of a different
statistical analysis (Maddala, 1984; Tobin, 1958). Even though there are no serious
violations of the equal variances assumption in IUR, it is still helpful to conduct a Tobit
analysis on the same data in order to validate the regression results. Therefore, a Tobit
estimates were conducted on the final model derived from the backward regression. The
results of the Tobit analysis are presented in Table 5-10.
The Tobit estimates presented in Table 5-10 show that all predictors in the final
model remained statistically significant. The Chi-square value was highest for the
democratization variable, followed by Muslim religion, GNP per capita, and teledensity.
The relative magnitude of the Chi-square values resembles closely that of the t-statistics
in the multiple regression analysis while the significance of the predictor variables is
Table 5-10. Tobit estimates for final model
a, b, c
a. The table reports Tobit estimates with Chi-square values and probability levels in last column.
b. Dependent Variable: Log Internet users
As expected, the regression coefficients and the coefficients in the Tobit analysis
remained almost identical. However, the Tobit table reveals that the standard errors for
the coefficient estimates went down, including both the intercept and the other four
variables. The probability levels also went down, compared with the p-values of the final
regression table, indicating even more strongly that the estimates were not produced just
be chance. The Log likelihood for normal in the Tobit estimation was -.465.
The results of the multiple regression analysis, in sum, showed substantial evidence
for the proposed relationships among the variables. Three of the hypotheses were
supported, one was partially supported, and three were not supported. Significantly, more
than 92 percent of the variation in Internet use was explained using the complete
multivariate model and about 90 percent was explained by the reduced final model. The
next chapter presents a discussion of the regression results in view of the literature
review. Suggestions for future research are offered in the final chapter.
Internet growth around the globe has been phenomenal. It has been argued that the
speed of Internet adoption has been unprecedented in the history of communication
technology. According to Nua's September 2002 data, the current worldwide Internet
population is 605.60 million (Nua, 2002). Projections for 2004 estimate an increase to
709. 1 million Internet users around the world (Cyberatlas, 2002). The Internet offers
potential benefits to nations worldwide in the areas of the political development,
economic progress, technological advancement, health and education. The World
Development Report titled Knowledge for Development noted that today even the
"remotest village has the possibility of tapping a global store of knowledge beyond the
dreams of anyone living a century ago, and more quickly and cheaply than anyone
imagined possible only a few decades ago" (World Bank, 1999, iii).
The Internet has changed world communications and the development of nations,
yet research on what affects its growth internationally remains inconsistent and
inconclusive. The main goal of this study was to identify the most significant predictors
of Internet adoption in the post-communist countries. A brief summary of the results is
presented below, followed by a discussion of the descriptive statistics and the hypothesis
This dissertation proposed and tested a five-dimensional theoretical framework to
explain the variations in Internet use across the post-communist countries. Three factors
emerged as critically important: economic, political, and technology/infrastructure
factors. Cultural factors seemed to exert some impact, but the results were inconclusive.
These findings suggest that the traditional country-level indicators of economic wealth
and technological infrastructure are just as important, and maybe more so, in today's
digital age. They serve as strong determinants of Internet use in the countries of Eastern
Europe and the former Soviet Union. But democratization level emerges as even more
important than income and infrastructure in that region. More democratic societies, which
offer greater freedoms to their citizens, are likely to encourage more extensive use of the
global network of networks.
The post-communist countries vary widely in Internet use. Looking closely at these
28 countries, there is a noticeable gap, which can be considered a regional digital divide.
But what exactly leads to such higher Internet use in one post-communist country and
such lower Internet use in another? Some have argued that the answer is national wealth.
If this were simply a question of income, though, then Slovenia should be the Internet
leader in the region. However, this is not the case. If it were just an infrastructure issue,
on the other hand, we should have observed phone-deprived Albania as the Internet
laggard in the region. Yet this is not the case either. The process of Internet adoption is a
complicated phenomenon influenced by a multitude of factors. The results of this study
shed some light on the macro-level indicators that drive Internet growth in the post-
communist countries in particular. The findings may be applicable to other regions and
other countries around the globe. Yet it should be underscored once more that this
analysis focused only on the 28 nations of Eastern Europe, the former Soviet Union, and
Mongolia. Therefore, applying the five-dimensional theoretical framework proposed here
should be made within the regional context.
The results of this study clearly identified four specific factors as critical
determinants of country-level Internet adoption: democratization, teledensity, national
income, and Muslim religion. All of these factors except Muslim religion have a positive
impact on Internet diffusion levels. Being a predominantly Muslim country, however,
seems to exert a negative impact on Internet diffusion. This can be an indication of strong
cultural differences associated with countries where Islam is the dominant religion.
Another possible explanation for the negative relationship could be that, incidentally, the
Muslim countries in the region have been historically in the "backyard" of the former
Soviet bloc. The Central Asia republics of the former Soviet Union in particular have had
less developed infrastructure and lower political autonomy in the past. Thus, the negative
relationship between Muslim religion and Internet use could be only a reflection of long-
lasting historic legacies.
The rest of this chapter goes deeper into interpreting the presence and absence of
relationships when testing the five-dimensional theoretical framework. First, we briefly
look at the descriptives and the meaning of Hypothesis 1. Next, we turn to each of the six
initial predictors and attempt to explain their influence on Internet use in the post-
Discussion of Descriptive Analysis
This section briefly reviews the range of variations among the post-communist
countries across the six variables of interest: national income, democratization,
telecommunications policy, teledensity, education, and religion.
The majority of the 28 post-communist countries are lower middle income, a few
are low income, and only six are classified by the World Bank as upper middle income.
Only one country-Slovenia— is a high income country according to the World Bank
classification. Most of the countries in the region, however, remain in the lower middle
Naturally, the countries in the region differ widely in their level of democratization.
None of the post-communist countries is ranked as completely free or democratic,
compared with the rest of the world. Nine countries are ranked as mostly free and can be
considered aspiring democracies. Only one country is absolutely undemocratic-
Turkmenistan. As expected, that country has lower Internet use than most; it ranks #26
among the 28 countries. The rest of the post-communist countries are in the middle range
of the civil liberties ranking, with some having made more progress than others. Belarus
is an example of a country where political and civil freedoms have been restricted by the
government. It ranks #23 among the group.
Telecommunications privatization in the region as a whole has been slow.
Privatization of the major telecommunications operator was successful in 2 out of 5 cases
only. Several countries including Estonia and Hungary have successfully privatized their
PTTs. The majority of countries, however, are lagging behind in the telecommunications
Telephone penetration in the post-communist countries is relatively high compared
with Africa, for instance, but lower than Western European levels. The teledensity across
the region also varies widely: the country with highest teledensity -Estonia— leads the
country with lowest teledensity- Albania-by a factor of 16. Mobile phone penetration
has increased throughout the region, but remains much lower than landline telephone use.
Yet traditional residential phones remain outdated in terms of connection quality.
The post-communist countries have been well known for their advanced
educational system. Traditionally, college attendance in the region has been high. Not
surprisingly, there is relatively little variation in terms of tertiary education ratio in the
region. Compared to other regions, the post-communist countries rank higher in relative
terms. Their mean tertiary education ratio is approximately 28 compared to tertiary
percentage enrollment of 1 in Angola and Haiti, 1 1 in Iraq, 15 in Brazil, and 16 in
Mexico. The education level of Western European nations, however, is somewhat higher:
62 for Norway, 5 1 for France, and 45 for Denmark.
The most common religion in the region is Eastern Orthodox religion. The other
two major groups are Western Christians and Muslims. Western Christianity includes
Protestants and Catholics. It is interesting to note that most of the former Soviet republics
located in Central Asia are Muslim countries. Compared to other regions, the post-
communist countries are more uniform in terms of religious composition. A clear
dominant religion exists and the population is highly homogeneous for the countries in
this group unlike the United States, for example, where different ethnic minorities and
various religious affiliations are abundant.
Growth of Internet Use
The Internet is one of the fastest growing communication technologies (WIPO,
2001). It has been adopted globally at an unprecedented pace. Yet, as earlier chapters of
this dissertation noted, many world regions are still behind in Internet use compared with
the United States and Western Europe (Daly, 1999; Dasgupta et al., 2001; World Bank,
1999, 2000). Even though Jupiter Research projects that by 2005, the U.S. share of the
Internet population will drop to 24 percent (Jupiter Research, 2001a), there is still an
imbalance between Internet-rich and Internet-poor countries. Indeed, there are huge
inequalities in Internet access and use around the world (Norris, 2000).
Some striking figures help illustrate the magnitude of this digital divide. For
example, Dasgupta et al. (2001) show that in 2000, 90 percent of the world's Internet
subscribers came from countries with 15 percent of the world population (Dasgupta, et al.
2001). Developing countries account for only 26 percent of all Internet users, with only 2
percent of the population in these countries being online (ITU, 2001). These numbers
illustrate the severe inequality in terms of Internet access across countries worldwide.
Sub-Saharan Africa is one region where the adoption of Information and
Communication Technologies (ICTs) has been much slower. Eastern Europe is in the
middle of the road, faster in adoption of the Internet than African countries, but slower
than Western European countries. The post-communist countries seem to parallel Internet
adoption and penetration in Central and Latin American countries.
Still, the number of Internet users in the post-communist countries remains slim
compared to that in the United States. Howard et al. (2001) report that the Internet has
become a vital part of the lives of Americans. Data show that 87 percent of the American
population uses email and 33 percent uses the Internet to retrieve information regularly.
In addition, the study reports that 21 percent of Americans use the Internet for major life
activities such as doing research online about health care and jobs and 9 percent also
make transactions online (Howard et al., 2001).
One important finding of a 2000 survey is that the younger Internet users are "more
likely to . . . gather most kinds of information, and to perform financial transactions
online" (Howard et al., 2001, 390). They also conclude that there are "a variety of
demographic factors that affect people's use of the Internet, including gender, age,
education, income, race and ethnicity. But the most useful predictors of the activities that
users enjoy online are their length of experience with the Internet and their frequency of
logging on from home" (Howard et al., 2001, 403). Clearly, the length and frequency of
Internet use are important predictors of online activities (Howard et al., 2001).
Demographic variables also continue to be important predictors of Internet use.
The size of the American online population has increased tremendously over a
short period of time. Katz et al. (2001) show that the number of users in the U.S.
increased from 8 percent in 1995 to 65 percent of the population in 2000 (Katz et al.,
2001). In terms of composition, they observe that the number of users age 40 and older
increased over time. Among a group of demographic predictors, income and education
emerge as the most salient factors affecting people's awareness of the Internet (Katz et
al., 2001). They conclude that the digital divide within the USA is shrinking.
Hypothesis 1 of this dissertation addressed the question whether the average
number of Internet users in the post-communist countries increased significantly from
1995 to 1999. The statistical comparison clearly indicated that, indeed, there had been a
significant increase throughout the region.
The number of Internet users in the post-communist countries did increase
significantly during the five-year period from 1995 to 1999. Even though this hypothesis
only tested the trend for 19 countries (no data for the other nine were available for 1995),
it is fair to predict that the growth in the rest of these countries will follow a similar trend.
In fact, looking at the descriptive statistics for Internet users in the region, we expect that
the countries lagging further behind will exhibit faster adoption and increase use at a
higher speed. This, of course, can be extrapolated not only from the descriptives, but
from the diffusion S-curve, which postulates that after a steep increase in the adoption of
any technology, the diffusion speed decreases and adoption levels off.
Compared to the rest of the world, the post-communist countries fall in the middle
range in terms of Internet adoption. On average, about seven percent of the population in
the post-communist countries is online. Internet indicators reveal an unexpectedly strong
disparity in usage across the countries. The Economist found that Hungary, for instance,
is ranked much higher than Ukraine in e-readiness (The Internet's new borders, 2001).
The Czech Republic and Slovenia are also more advanced than Albania and Azerbaijan
in terms of Internet access, use, online commerce and other online activities. In 1999,
Estonia, for example, had 175 Internet hosts per 10,000 people while Albania had only
0.24. Data on Internet penetration in Hungary show that 7.1 percent of the population was
online in October 2000 (Minges, 2001). On average, about seven percent of the
population in the post-communist countries was online as of 2000. This number is
growing rapidly (CDT, 2000; The Internet's new borders, 2001; ITU, 1999).
There seems to be a digital divide in the region, with Estonia and the Central
European countries at the top and the Central Asia former Soviet republics at the bottom.
The three countries that are leaders in Internet adoption are Estonia, the Slovak Republic,
and Slovenia. The lowest eight countries are Albania, Azerbaijan, Belarus, Bosnia and
Herzegovina, Mongolia, Turkmenistan, Tajikistan, and Uzbekistan. There is clearly a
wide disparity in Internet use across the post-communist countries. The difference in
Internet users per 10,000 people between the top and the bottom was more than 400 times
The Internet has become important in some of the countries, where specific uses
and projects have emerged. The Internet made possible, for instance, the development of
medical centers within the region. One project links a major hospital in Ukraine with
doctors in Finland for real-time medical support.
Discussion of Hypotheses 2 through 6
This section provides an interpretation of the results of the multiple regression
analysis used to test Hypothesis 2 through Hypothesis 6.
The most commonly used predictor of per capita Internet use is probably national
income, measured as Purchasing Power Parity Gross National Product per capita (PPP
GNPP) in this study. The second hypothesis stated that higher per capita income would
contribute to a higher number of Internet users per capita. The regression results strongly
supported this hypothesis.
The positive impact of national income on Internet use may come as no surprise to
some scholars. Previous studies on Internet diffusion across countries have shown the
importance of economic development for ICT growth. Thus, the results of this study are
in agreement with Arnum and Conti (1998), Bazar and Boalch (1997), Elie (1998), and
A few previous studies, however, have shown no significance of macro-level
economic factors. Income per capita was insignificant in a study conducted by the World
Bank, for example (Dasgupta et al., 2001). The results of this research refute such studies
and suggest that Internet disparities across the post-communist countries are more than
just a reflection of infrastructure inequalities, as some studies suggest. The digital divide
in the region is due, at least in part, to the economic inequalities between the countries.
Future studies of cross-country Internet diffusion should incorporate national income as
an important determinant, but take into consideration that it is not the only determinant
that plays a role in the process.
It was expected that GNP per capita might turn out to be the variable with strongest
explanatory power in the final multiple regression model. Some studies have shown that
national income level is the strongest predictor of Internet adoption (Beilock &
Dimitrova, 2003; Elie, 1998; Hargittai, 1999). This was not the case in our findings. One
reason for the relatively lower significance of national income among the predictor
variables may be partly due to the multicollinearity between GNP and teledensity. The
implication of such multicollinearity is that these two variables, if entered together in the
model, may lower each other's significance.
In this study, national income was the third most significant determinant of Internet
diffusion, following democratization and teledensity. As Internet connection prices go
down relative to average wage, the income factor may be further reduced in its
importance in the post-communist world. This is supported by a comparison of Western
European nations where income did not appear significant as a predictor of Internet usage
rates (Beilock & Dimitrova, 2003). It may be expected that the importance of income
would wane over time. It should be noted, however, that national income level will
remain an issue in countries where the hourly rate for Internet connection equals or
exceeds one percent of the monthly salary.
Level of democratization was used as a political factor in the five-dimensional
analytical framework. Compared with income and infrastructure factors, democratization
is not incorporated as frequently in diffusion studies. It was considered an important
factor in this analytical framework, however. This study predicted a positive relationship
between the level of civil liberties and Internet adoption levels. The results of the
multiple regression showed that the higher the level of civil liberties in the country, the
higher the number of Internet users per capita.
This finding supports the argument that political factors are critical for Internet
development. Civil liberties are an indicator of the level of democratization. The process
of democratization has been particularly important in the region. In fact, the word
"democratization" is often used synonymously with the word "transition" in the post-
communist world. Democratization as a broad term also refers to the opening of society
to the rest of the world and not just to the political and civil freedoms allowed in the
Norris (2000, 2001) observed a recurring trend that established democracies around
the globe tended to have higher Internet penetration. Bivariate correlations clearly
showed that higher level of democratization and higher Internet use were positively
related (Norris, 2001). Yet, Norris (2001) found no statistical significance of the
democratization variable in her world model. In this study, however, democratization
emerged as the most influential factor in the region of the post-communist countries. This
could be partially due to the specificity of the region examined in this study.
Civil liberties appear to be critical for Internet adoption in the post-communist
countries. The civil liberties composite variable used here includes several areas. One is
the freedom of expression and belief. The existence of free and independent media then is
positively related to Internet growth. Another aspect of civil liberties is the freedom of
assembly and demonstration, in other words, freedom to protest openly. It may not be far-
fetched to argue that countries which do not allow freedom of assembly are likely to
restrict Internet use as well. In addition, an independent judicial system, respect for
human rights, personal autonomy and economic rights are also important civil liberties.
We infer that countries that restrict such rights are likely to have lower Internet
These conclusions are supported by Rodriguez and Wilson's (2000) study, in which
they underscore the significance of political freedoms and civil liberties for the
technological progress of developing countries. They argue that the existence of a climate
of democratic freedoms not only facilitates, but also is a requirement for the faster adoption
of ICTs (Rodriguez & Wilson, 2000). The existence of civil liberties as a necessary
condition for Internet adoption has been overlooked in the literature so far.
It is important to note that there may be exceptions to the rule. Some countries with
undemocratic governments have very high Internet use. Singapore and China are two
examples of that. While these countries were not examined in this study, it is plausible
that in both cases high Internet use may be related to systematic government policies to
encourage use. In the case of Singapore, it is important to note that the country has the
lowest Uncertainty Avoidance Index (UAI) score in the world (Hofstede, 2001).
Therefore, some cultural factors may dominate over the level of democratization in
society in certain cases.
Many scholars have underscored the significance of government policies when
discussing Internet adoption. Petrazzini and Guerrero (2000), for example, note the
importance of telecommunications policy in the Latin American countries. Tanner (1999)
discusses the significance of telecommunications policy in regard to Western Europe.
Sallai (2000) and Wolcott et al. (2001) also have noted the importance of policies in the
telecommunications sector for the growth of Internet use.
The five-dimensional analytical model proposed that a longer period of
telecommunications privatization would lead to a higher number of Internet users across
countries. Privatization was not a significant predictor in this study, however. This
finding is in accordance with the Kiiski and Pohjola (2001) study, which finds no effect
of telecommunications policy on country-level Internet use. However, we need to be
cautious in interpreting that telecommunications privatization has an effect on national
One possible explanation for the lack of significance of the telecom privatization
variable may be simply the distribution of the data points. In other words, the lack of
importance may be largely due to the fact that many of the values of the policy variable
equal zero as of the end of 1999. If that is indeed the case, it is likely that the effects of
telecom privatization will be visible in the long run, and that telecommunications policy
will show increased importance over time.
Previous research again in the Western European countries found a strong
explanatory power in policies regarding the telecommunications sector. Hargittai's
(1999) research on Internet connectivity among the Organisation for Economic
Cooperation and Development (OECD) countries is one of the important studies that
show a significant relationship between Internet use and telecom policy. She showed that
countries that have allowed free competition, or even some degree of competition, have
higher Internet penetration than countries with telecom monopolies (Hargittai, 1999).
Even though the results of this study do not support Hargittai's findings, it is, again,
likely to see stronger effects of telecommunications policy on Internet use in the near
The apparent contradiction between Hargittai's conclusion and the findings of this
research may be explained by the fact that each study measured telecommunications
policy in a different way. Hargittai (1999) used the level of competition (monopoly,
duopoly, competitive market) in the telecommunications sector as a predictor. That
predictor reflects one aspect of telecom policy—demonopolization. This study, however,
modeled telecom policy by a privatization variable. It appears that length of privatization
may be insignificant for Internet adoption in contrast to level of competition in the
telecom market. We need to distinguish between these two different aspects of
telecommunications reform, but also bear in mind that different policy dimensions may
influence Internet use differently in different regions. Hargittai's (1999) study focused on
OECD countries (Western European nations) while this study examined Eastern
European countries and former Soviet republics.
Still, telecom market privatization may play an important role in the post-
communist world's technological future. A number of previous studies have suggested
that it is critical to include the level of privatization when studying Internet adoption in
the post-communist countries (Ellis, 1999; Fish, 1998; Gulyas, 1998; Kuentzel et al.,
2000; Maddock, 1997; Papir & Oleszak, 2000; UNDP, 1999). Even though the effects of
privatization in general and the main telecommunications operator in particular may be
yet invisible, their long-term impact in the transition countries needs to be followed.
Future studies should also consider other measures of telecommunications policy such as
level of competition in the telecommunications market in addition to length of
Technological infrastructure has often been described as one of the barriers to
increased Internet diffusion. When there are more telephones in New York City than in
several African countries combined, the existence of strong cross-country Internet
discrepancies is hardly a surprise. This study tested the relationship between teledensity
and Internet use per capita. Even though the relationship between the two variables was
positive, it was not significant at first in the complete regression model. However,
subsequent analysis showed that telephone infrastructure emerged as the second most
important factor in the Internet diffusion process.
The teledensity variable, even though insignificant in the complete model,
remained in the final model after backward regression was conducted. There are at least
two possible reasons for this phenomenon. The most likely reason may be that there is
some indirect relationship between teledensity and telecommunications privatization,
both of which are in the full regression model. Conceivably, countries which privatized
their telecom operator earlier have higher teledensity to begin with, which makes the
effect of both variables disappear. Therefore, teledensity becomes significant once the
less influential privatization variable is removed from the regression model.
Another possible explanation could be the high multicollinearity between GNP and
teledensity discussed earlier in this study. In cases of such high correlation between
explanatory variables, it is common to see that one variable increases its significance at
the expense of the other one and vice versa. Additionally, even though teledensity is not
statistically significant in the complete model, it still contributes to its overall explanatory
Since existing infrastructure significantly affects Internet adoption at the country
level (Arnum & Conti, 1998; Bazar & Boalch, 1997; Elie, 1998; Gulyas, 1998; Hargittai,
1999; Lin, 1998; Sadowsky, 1993), variations in Internet use are to be expected between
countries with different levels of infrastructure development. This study looked
specifically at telephone infrastructure. The relative degree of difference between
countries in the post-communist world, however, may be much smaller than the
differences that exist between Rwanda and France or Morocco and Finland, for example.
Telecommunication infrastructure varies widely across continents, regions, and countries
(Daly, 1999; ITU, 2000). The variations are not as stark within the former Soviet bloc
region, however. Therefore, we expect that the importance of pre-existing telephone
networks may be higher when comparing all countries in the world. Beilock and
Dimitrova (2003), among others, have shown that teledensity is a significant determinant
of adoption, examining a world model of 105 countries. Telephone infrastructure then
appears to be critical for Internet adoption on a global scale.
It is important to note that this dissertation is one of the first studies to include
mobile phones and not just residential phones as part of the technology infrastructure.
This is an innovative approach, which may be useful in future research because the
importance of mobile telephone infrastructure is expected to grow over time. The only
drawback of this approach is that the individual effects on residential versus mobile
phones cannot be isolated. Thus, we can only conclude that both residential and mobile
phones are significant in the Internet adoption process.
Previous research, for the most part, has shown that educational level is an
important determinant of Internet use. The results of this regression analysis, however,
did not support this proposition. In fact, higher education seemed to be related to lower
Internet use in the region of interest. Even though the results for the education predictor
were not statistically significant, we need to address not only the lack of significance, but
also the negative sign of the correlation coefficient between college education and
It is important to note that the post-communist countries offer good quality
education—both at the high school and college level. Even though the 1999 Human
Development Report underscores that education in the region as a whole has recently
deteriorated, both in terms of quantity and quality (UNDP, 1999), the basic educational
level of the general population remains high. Typically, more educated people tend to be
earlier adopters of technology (Rogers, 1995). Yet we found a negative relationship
between educational level and Internet use.
Does education matter? The results of this study are inconclusive. The tertiary
enrollment ratio used here as a measure of educational achievement was not statistically
significant. Even more surprisingly, the regression coefficient for that factor was
negative. This seems contradictory to previous studies that found a positive relationship
between education and Internet use. There are at least two possible explanations for this
First, it seems that younger people tend to be heavier Internet users (Howard et al.,
2001). This is particularly true in the post-communist countries (Kouznetsov & Bourtsev,
1996). Rose (2002) notes that in Russia, for instance, more than one-sixth of the people
under age 29 are Internet users. In other words, age is inversely related to use. Older
people in general are slower in the adoption of new technologies, such as the Internet.
Yet, older people tend to be more educated. Being older and more mature means being
more educated than the younger population, but also, arguably, less likely to use the
Internet. Therefore, including education without controlling for age may produce
Additionally, basic use of the Internet may be simple enough for people with less
than a college degree. How much knowledge does one need to use the Web? The World
Wide Web combines text and graphics, and learning the main navigation tools today may
require little instruction. Email, which is the most popular use of the Internet around the
world, tends to be user-friendly and easy to use. A better education variable to be used
could be a specific measure of ICT skills. Finally, UAI tends to be lower among younger
people. Future research should try to incorporate UAI into the education variable in order
to avoid possible interaction.
One of the few studies that supports the results of the current research was
conducted by Kiiski and Pohjola (2001). These researchers also found no statistical
significance of the education variable in their analysis of cross-country Internet
penetration. It is suggested here that educational level is not an important determinant of
Internet use during the early adoption stage. However, it may become more significant at
later stages of the Internet adoption process.
The study also hypothesized that differences in dominant national religion would
affect the number of Internet users per capita. The results of the regression analysis
showed partial support for this hypothesis. It seemed that Muslim religion had a
significant effect on Internet use in the post-communist countries, but Western
Christianity did not.
Religion was incorporated in this study as a cultural measure. The dominant
religion in the country was measured as Eastern Orthodox Christianity (the most common
religion in the group), Western Christianity (Catholic or Protestant), and Islam. It was
argued that these three religions were substantially different from one another (Fish,
1998), and that religion was a reflection of substantial cultural differences across the
countries (Hofstede, 2001).
Based on the two dummy variables employed in the study, some interesting
findings emerged. First, Western Christianity did not seem to have a statistically
significant effect on the level of Internet adoption in the country. Its regression
coefficient, however, was positive. In contrast, the coefficient for Muslim religion was
negative and statistically significant. This clearly indicated a negative correlation
between Internet use and Muslim religion. The causal relationship between the two
variables, however, is not so clear.
The following countries from the population of interest are predominantly Muslim:
Albania, Azerbaijan, Bosnia and Herzegovina, Kazakhstan, Kyrgyzstan, Tajikistan,
Turkmenistan, and Uzbekistan. Incidentally, the Muslim countries in the sample share
some historic legacies. Six of them are former Soviet republics, five of which are located
in Central Asia, and one— Azerbaijan— in the Caucuses. Arguably, these countries
received less technological support and less infrastructure investment while members of
the Soviet Union. In fact, the telephone infrastructure in some of the Central Asia
republics is still quite poor compared with 21 st century standards. Therefore, it is
conceivable that Muslim religion was significant in this study because it captured historic
inequalities in the realm of technology and national infrastructure.
To explore the question whether religion serves as a substitute for infrastructure in
the multiple regression, a model without the teledensity variable was tested. We
performed this regression with the expectation that Western religion would remain
significant when the teledensity variable was not included if the first explanation
suggested above was true. However, this was not the case. Muslim religion was
significant and negatively related to Internet use whereas Western religion was removed
from the model even when teledensity was excluded. It may be conceivable that Western
religion is positively related to democratization and educational level in addition to
teledensity. But it seems more likely that perhaps religion parallels broader cultural
differences among the countries of interest.
There is another possible explanation for the negative relationship between
predominantly Muslim societies and per capita Internet use. It is possible that religion is
correlated with some other cultural factor that is negatively related to Internet adoption.
Thus, religion may have a spurious relationship with Internet use and is most likely an
indication of some strong cultural differences between Muslim and Christian societies.
One such cultural factor could be, for instance, overall openness of society to new
technologies. This is similar to Hofstede's dimension of uncertainty avoidance (Hofstede,
Hofstede developed four dimensions of culture: power distance, uncertainty
avoidance, individualism and collectivism, and masculinity and femininity. The
dimension which relates to the adoption of innovations in general and the Internet in
particular is the uncertainty avoidance index (UAI) (Hofstede, 2001). Hofstede (2001,
161) defined uncertainty avoidance as "the extent to which the members of a culture feel
threatened by uncertain or unknown situations." Research shows that UAI is negatively
correlated with the adoption of new media and use of the Internet (Hofstede, 2001).
Dominant religion seems to reinforce the values that led to its adoption in a
particular country. Also it is important to note that religion is listed as one of the three
major ways in which society copes with uncertainty. High uncertainty avoidance
countries resist innovations and are more resistant to changes in general. In contrast, "low
UIA countries tend to have more open-minded mentality, in searching for information
and in accessibility to innovation" (Hofstede, 2001, 170). Religion may not cause
uncertainty avoidance per se— but both may have a common cause. Hofstede notes that
"an established religion reinforces the values that led to its adoption, however,
confirming either strong or weak uncertainty avoidance" (2001, 177).
For instance, two Muslim societies— Turkey and Pakistan-are ranked very high in
UAI. This suggests these societies as a whole will be less receptive to new technologies.
At the other end of the spectrum are Singapore (with the lowest UAI score), Denmark
However, no data exist on the ranking of the post-communist countries according to Hofstede's cultural
dimensions, with the exception of former Yugoslavia. Hofstede also notes that the uncertainty avoidance
ranking of Yugoslavia is not highly reliable. Today Hofstede's questions are included in the European
Media and Marketing Survey that is administered in the member countries of the European Union.
and Sweden. These countries have lower uncertainty avoidance, and anecdotal data show
that they are among the leaders in Internet use in the world.
In either case, this study partially supports the idea that culture plays a major role in
adopting technological innovations. However, culture is a very complex concept, as many
authors have noted (DiMaggio, 1997; Hofstede, 1980, 2001; Sondergaard, 1994; Tayeb,
Therefore, it is difficult to capture all dimensions of culture. Our observations and
data analysis showed that, controlling for income and infrastructure as well as for other
factors, Muslim religion was negatively related to Internet use whereas Western
Christianity had no effect. It would be necessary to test the effects of religion on a
worldwide model before making conclusive remarks in this regard. In addition, it would
be desirable to include other cultural variables in future studies of Internet diffusion,
when data permit.
Refined Conceptual Framework
This study proposed a five-dimensional theoretical framework for explaining cross-
country adoption of the Internet. The overall conceptual model included economic,
political climate and policy, infrastructure, audience, and cultural factors as determinants
of adoption. The study showed that three of these factors-namely economic, political,
and infrastructural-play an important role in the Internet diffusion process in the post-
communist countries. There was partial support for the influence of cultural factors, as
revealed by the effect of religion. The policy variable used here— telecommunications
privatization-did not appear significant. Finally, the importance of audience
characteristics measured by education in particular was not supported.
The four specific variables that remained highly significant in this study were per
capita income, democratization, teledensity, and Muslim religion. These four predictors
explained a large amount of variation in Internet use in the post-communist countries. As
expected, higher income, democratization, and teledensity led to higher Internet adoption
levels. These highly significant predictors of Internet adoption are likely to remain
important in the future. Being a predominantly Muslim country seemed to affect Internet
use negatively whereas being a predominantly Protestant or Catholic country had no
effect. This partial support for the influence of religion points to the potential impact of
certain cultural predispositions. However, better measures of cultural differences need to
be developed in order to gather more evidence of their influence on Internet adoption.
Surprisingly, the telecommunications policy variable was not significant in this
study. However, this finding may have been driven by the specifics of the data and the
brief period of time over privatization was examined. It is expected that the privatization
policies would show an effect in a time-series analysis over an extended period.
It is also conceivable that a different policy measure— for instance, telecom market
competition— may be a better determinant of Internet adoption. This study measured only
one aspect of telecommunications policy— privatization of the national
telecommunications operator. Privatization by itself may not be sufficient for bring about
demonopolization in the telecom sector. Both privatization and demonopolization are
considered important aspects of telecom liberalization in general.
Finally, we found educational level to be irrelevant for Internet adoption in the
post-communist countries. It is conceivable that this is an artifact of the relatively small
variation of the education variable across these countries. Another possible reason is that
the early adopters of any technology tend to be highly educated. Therefore, it is likely
that the early adopters in each of the post-communist countries are the highly educated
elite to begin with, so looking the average number or the whole population may be
misleading. If this is indeed the case, the influence of education is likely to be more
visible at later stages of Internet adoption when a higher percentage of the population
joins the online community.
The most convincing argument for the lack of significance of education seems to be
its inverse relationship with age. Younger people tend to have lower education but seem
to be the more likely Internet users. It is also important to point out another possibility:
overall educational level may be misleading because it is not education per se that we
want to capture, but rather the specifically Internet or computer skills of the population.
Some clarification is necessary regarding the role of audience factors in the Internet
adoption process. An important audience variable is English language proficiency.
Considering that the majority of Internet content today is in English and that many
operating systems and software programs (such as Microsoft Windows and Internet
Explorer) are available only in English in the countries of interest, this variable is needed
as a determinant of adoption, even though data for it were not available at this stage.
Thus, future studies should include English language proficiency and test for its effect on
national Internet usage levels.
This dissertation examined only the effects of macro-level determinants. Of course,
there are other factors that affect diffusion at the individual and institutional level. Norris
(2001) conceptualized Internet diffusion to be studied at the following three levels:
macro, meso, and micro level. An extension of this research should zoom in on the
countries of interest and determine the effect of meso- and micro-level factors. These,
again, may become more relevant at later stages of adoption.
In sum, the results of this study suggest that even though economic factors are
important for Internet adoption in the post-communist countries, infrastructure, political
and cultural factors do matter as well. The findings of this dissertation have important
implications not only for the post-communist counties, but for nations across the globe.
These theoretical and practical implications are discussed in the final chapter.
The global digital divide has become a concern for policymakers at both the
national and international level. Kofi Annan, the United Nations Secretary-General,
warned about the possibility that the poor of the world could be excluded from the
information revolution: "These people are deprived of many things: work, shelter, health
care, and drinking water. But to be isolated from the major telecommunication services
now is an equally grave deprivation, and chances to remedy this situation are shrinking"
(Annan, 1999). The growing digital divide is even more disconcerting because the gap
between the developing and the developed world in information and communication
technology adoption is greater than the traditional gaps of income and welfare (Daly,
1999; Rodriguez & Wilson, 2000). As explained below, the results of this study can be
used by the lesser connected countries to improve their Internet connectivity and thus
help bridge the gap between the Internet-rich and Internet-poor world.
This study developed a five-dimensional theoretical framework to explain the
variations of Internet use in the post-communist countries. Based on the findings of the
study, we conclude that economic, political, and technology/infrastructure factors
positively affect Internet use in the region. Cultural factors also seem to have an impact,
with Muslim religion adversely affecting Internet use. This negative relationship,
however, may be, coincidentally, a reflection of historic inequalities between the post-
Using multiple regression to test the five-dimensional framework proposed in
Chapter 3, it was determined that democratization, telephone infrastructure, and national
wealth were the three most important factors that affect Internet use in the post-
communist countries. Other determinants suggested in the literature did not appear to
influence Internet use in the region. Having predominantly Western Christian religion
seemed unrelated to Internet adoption. National policy— and length of
telecommunications privatization in particular— did not appear significant in the analysis
either. Additionally, college education was not found to be related to Internet use in the
post-communist countries. Contrary to expectations, higher level of education did not
predict higher Internet use in these countries.
Three of the five factors proposed in the Internet diffusion framework were found
to be important. On theoretical grounds, we can derive a refined conceptual framework
based on the empirical results of the study. The new framework includes (1) economic;
(2) political; and (3) infrastructure factors. Cultural factors may be important as well, as
evidenced by the significance of Muslim religion. Including religion or some other
cultural measure is recommended as a fourth component. A measure similar to
Hofstede's uncertainty avoidance may be more revealing in capturing cultural differences
across nations than religion. Finally, it was observed that national income and telephone
infrastructure are highly correlated, so looking at the effects of one and disregarding the
other would be a mistake.
The variables which were not significant in this study were telecommunications
privatization and educational level. It is recommended to find better proxies for the
effects of telecommunications policy and education, respectively. We conclude that the
former should also be measured by telecommunications demonopolization or level of
competition in the telecom sector and the latter should capture not education per se but
specifically Internet training and ICT skills.
Finally, it is expected that the effects of income and infrastructure will wane over
time. On the other hand, the weight of cultural factors is likely to increase at the later
stages of the Internet diffusion process, as suggested by Beilock and Dimitrova (2003).
This research has important theoretical and practical implications. The findings of
this dissertation can be of value to at least three major constituencies: policymakers,
business companies, and the academic community.
The findings of this dissertation contribute to the academic research by (1) offering
a descriptive analysis of Internet use; (2) showing which factors affect Internet adoption
in the post-communist world; (3) developing a comprehensive model of Internet diffusion
that can be applied to other countries and regions of the world; and (4) extending
diffusion of innovations theory to the country level and showing which of its
generalizations apply to Internet adoption in the post-communist countries.
This dissertation fills a gap in the literature because it offers an extensive
description of Internet adoption in the post-communist countries. No previous studies
have examined Internet variations in the region in such detail. This study is one of the
first to show clearly the existence of a digital divide not only around the world, but also
among the post-communist countries.
Further, the dissertation identifies a specific set of factors that affect Internet
adoption in the post-communist world. Thus, it contributes to the body of literature on
cross-country Internet diffusion by shedding light on the determinants of the diffusion
process in a specific region of the world. The results clearly show that economic,
political, and infrastructure factors measured by per capita income, level of
democratization, and teledensity respectively, influence Internet use in the region. There
has been a void in the literature on Internet adoption in countries other than the
industrialized countries of the world, and our findings fill this void.
Probably the main theoretical contribution of the dissertation is that it develops a
more comprehensive model of cross-country Internet diffusion than any previously tested
models. The proposed five-dimensional model is broader and more accurate than earlier
bivariate models. In addition to the economic and infrastructure variables tested in
previous studies, it adds the variables of culture and democratization. This
comprehensive model of Internet diffusion also incorporates audience characteristics and
policy variables, which have been suggested in previous studies, but rarely included in a
multivariate analysis. Another advantage of this five-dimensional framework is that it can
be easily applied to other countries and regions of the world and not only to the post-
Finally, the dissertation used some of the generalizations of the diffusion of
innovations theory and applied those to the country level of adoption. Our results showed
that income and education as well as culture can be translated and used not only at the
individual but also at the societal level of adoption. However, a more appropriate
measure of education than college education may need to be developed when studying
Policymakers at the local, national, and international levels can use this study to
develop or modify policies aiming at accelerating Internet use in the former Soviet bloc
countries. The results of the study can be used as guidelines for Internet diffusion in other
countries as well. Governments, non-governmental organizations (NGOs), and
international organizations can benefit from the empirical results of this study, assuming
that these organizations have a vested interest in promoting Internet development in the
so-called transition economies of Eastern and Central Europe and the former Soviet
The findings of this research are particularly revenant to NGOs and international
organizations that are trying to understand and promote Internet adoption around the
world. As Rodriguez and Wilson (2000) note, multilateral and bilateral support for
information and communications technology activities in developing countries is warranted
because of the increasing global technological gap. Organizations such as the Open Society
Foundation, the International Telecommunication Union, the United Nations, USAID,
and the World Bank may benefit from the results of this dissertation and use them as
guidelines for their international Internet-related projects.
The study examined Internet adoption in the post-communist countries. The results
identified three main factors— democratization, infrastructure, and national wealth—that
facilitate Internet diffusion at the country level and suggested other factors that serve as
barriers. The results demonstrated not only the direction, but also the strength of
relationships between these factors.
The level of democratization emerged as the most important determinant of Internet
adoption relative to the other predictors in the framework. This finding may be deemed
relevant to policymakers because it shows that limiting democratic freedoms affects
Internet adoption negatively. On the other hand, encouraging free expression,
independent media, and freedom of assembly can lead to increased Internet penetration.
The findings of this dissertation also confirmed the importance of teledensity in the
process of Internet adoption at the global level. Teledensity emerged as the second most
important determinant after level of democratization. Organizations such as the World
Bank that deal with inequality around the world may base their future projects on the
knowledge that infrastructure gaps still serve as major barriers to Internet adoption, even
in the post-communist countries.
Consistent with previous research, national income was found to be significant for
country-level Internet adoption. It was the third most important predictor in this study.
One way to increase adoption then would be to increase the wealth of the country. This
finding has ramifications for national governments and loan granting agencies at the
national and international level.
The practical nature of this study also makes it useful to the business community.
Companies in Internet-related businesses that plan expansion to the post-communist
countries can find the descriptive results of the study useful. The study could help them
make strategic decisions on which country to enter first (e.g., Estonia would be a good
choice since it is the leader in Internet adoption in the region). In addition, this
dissertation can be helpful to local Internet companies and Internet service providers
(ISPs). The ISPs can adjust their pricing structures accordingly based on the finding that
economic factors serve as a barrier to Internet diffusion. Finally, the study has potential
utility to market research companies that measure and forecast Internet levels at the
international arena. Such companies can compare the speed of adoption in this region to
other regions in the world and possibly make some predictions when the Internet is likely
to reach universal penetration in the post-communist countries.
The potential benefits of Internet adoption and use were discussed at length earlier
in the manuscript. Probably the most amazing contribution that the Internet brings to
people and organizations worldwide is the quick, cheap, and seamless transfer of
knowledge. "Knowledge is critical for development, because everything we do depends
on knowledge" (World Bank, 1999, 16). The Internet provides a venue for any individual
to learn about a multitude of topics—from successful business competition strategies to
how to treat infant illnesses. Therefore, increased Internet access is likely to bring more
independence and knowledge to people of all walks of life.
Some studies have shown concern that the benefits of the Internet would be reaped
only by the highly developed nations and that less developed countries would gain less
from increased Internet use. Roller and Waverman (2001), for example, argue that
increases in telecommunications infrastructure lead to higher growth among more
economically developed countries, such as the OECD countries. The growth effects are
expected to be lower in less developed countries (Roller & Waverman, 2001). Daly
(1999) notes that, paradoxically, "those who have to get most Internet benefits, and those
who have the greatest needs have least access to the new technology." Further studies
should examine whether wealthier countries gain more benefits from new technology
adoption than less developed countries.
Even if this difference turns out to be true, there is still a bigger danger of leaving
the developing countries completely behind in the information revolution (Rodriguez &
Wilson, 2000). That can lead to further increase knowledge gaps (World Bank, 1999). On
the other hand, if developing nations institute policies and establish institutions to help
Internet adoption, then can catch up with the industrialized world (World Bank, 1999).
A report by the European Commission (2001) on the digital divide notes that the
benefits from increased ICT use may be more qualitative and are not as tangible.
Therefore, it is difficult to find hard numbers to support the argument of how beneficial
the Internet is. Some of the areas in which the Internet had been considered a positive
force are economic development, the political process, leapfrogging, and social life.
The Internet affects the economic situation in a country as it facilitates international
trade, lowers production and distribution costs, optimizes productivity within and
between companies. In terms of contributions to political sphere, the Internet allows
people to speak publicly and publish information to wider, global audiences. In addition
to serving as a regulation-free information source, the Internet provides an avenue for
political communication and mobilization.
Leapfrogging is the ability of countries less developed technologically to skip
generations of intermediate technology and adopt the latest one—in this case, the Internet.
The adoption of the latest technology is considered beneficial to the country. Arguably, it
serves to facilitate catch-up with more technologically advanced societies.
The Internet also affects social life. Online applications such as chat rooms and
email bring people of all nationalities and ethnic backgrounds closer together. These
online applications redefine social relationships within countries. The effect of virtual
communities are expected to increase in the future.
Increased Internet use also has certain ideological implications for the national
population. The Internet has been seen as a vehicle for transmission of a number of
ideologies. It allows more democratic communication to take place. Yet it also allows the
Taliban and other extremist groups to establish worldwide networks and to communicate
outside national borders.
This discussion leads to the question: Is technology neural? While it is beyond the
scope of this study to answer this question, it is important to differentiate between the
different functions of the Internet. There definitely some beneficial aspects: free and open
expression online, possibility for political participation, leveling off competition during
economic transitions, and finally social connectivity. However, some of the negative
spillover that the Internet can bring to nations are values of consumerism, individualism
and militarism. Therefore, it is imperative to have a clear distinction between the specific
uses of the global network of networks, which makes it difficult to predict the exact
implications of increased Internet use.
This study has several limitations. Problems with data availability made it
necessary to include proxy variables for a number of factors identified in the literature
review. In particular, cost of Internet connection and culture had to be measured with
proxy variables. English-language proficiency statistics were unavailable at the time of
this research. Lack of data is a common problem in the post-communist countries, where
no reliable regional comparative data sources exist (Fish, 1998).
In an ideal world, social science researchers will not have to deal with issues of
nonexistent or unreliable data. In reality, we choose the best data available to us when
conducting our research. However, it is important to understand the limitations of the
It is critical to assess the validity of the study to assume no systematic bias was
introduced into the results. Several forms of validity exist. The two broad types of
validity-internal and external validity— are reviewed below.
Internal validity refers to the question whether all variables, measures, statistical
procedures, and inferences were sound and robust. Babbie (1995) discusses four types of
internal validity: face, content, criterion, and construct validity. Face validity in a study is
gained by careful inspection of the concepts and their measures. Basically, it examines if
the measures and the results of the study seem appropriate at face value, which is true in
this research. All variables, test procedures, and findings of the study have high face
Content validity establishes that the measure covers the full range of concept's
meanings. Content validity is high for all variables used in this study except for culture.
As aforementioned, culture is a complex phenomenon, and its meaning involves more
than national religion. Lack of data for the uncertainty avoidance index of Hofstede
prevented us from using that arguably more appropriate measure of cultural differences
between societies. It was beyond the scope of this study to develop and estimate other
measures of culture.
Criterion or predictive validity refers to the questions whether scores obtained on
one measure can be accurately compared to those obtained with an already validated
measure of the same phenomenon. The predictive validity of our measures is high
because they are based on previous research and the results are consistent with earlier
Finally, construct validity applies to situations where there is no clear criterion for
validation purposes of a particular variable. In this case, a different measure related to
other measures already in theory is the use of teledensity as a combination of both
residential and mobile phones. Previous Internet adoption studies have incorporated only
residential phones as an infrastructure variable. The increasing significance of mobile
phones in the post-communist countries made it necessary to construct this new
The internal validity of the study overall is high because all measures were sound
and regression assumptions were met. One concern may be the relatively small number of
observations/countries used in this exploratory study. As explained earlier, however, the
study incorporates the population of interest-all 28 post-communist countries.
The dependent variable used to measure Internet adoption within a country was
Internet users per 10,000 as reported by the International Telecommunication Union. It is
very difficult to estimate how many people are using the Internet within a country
(Castells, 2001; Daly, 1999; Press, 1997; Zook, 2000, 2002). One of the challenges for
developing countries in particular is that people often use the Internet from cyber cafes,
which is unaccounted for by domain-based measures. Also, current Internet data
collection methodologies publish different figures of Internet use, which suggests that
they do not provide completely reliable data. The data on number of Internet users per
capita, therefore, need to be considered only an approximation of the actual number of
Internet users in the country.
Some of the predictor variables point to other issues that need to be addressed. The
Gross National Product variable (GNP) by definition reflects only officially reported
economic data. However, UNDP (1999) underscores that shadow economies, which
remain mostly unreported, are a substantial part of the national economies of the post-
communist countries. For example, the shadow economy as percentage of GDP was close
to 17 percent in Bulgaria and over 19 percent in Russia in 1996 (UNDP, 1999). The
results of this study, however, are based only on officially reported GNP statistics.
Somewhat surprisingly, education was not significant in this study. It appears that
the difference between Tajikistan and Slovenia in educational level may be less
pronounced than the difference between Tanzania and the Netherlands, for instance. The
lack of significance may be also a result of the way education was measured-as tertiary
It was unexpectedly difficult to locate English language proficiency data. Data on
percentage of high school students studying English exist, but only for the current and
future members (candidate countries) of the European Union. Therefore, English
language proficiency could not be included as an explanatory variable in this analysis. It
is suggested, however, that future studies of cross-country Internet diffusion should
incorporate English language as a determinant of Internet adoption.
Multicollinearity-the high correlation between several independent variables-was
an inherent problem in this research. Multicollinearity was highest between GNP and
teledensity, but the other variables also showed moderate degree of multicollinearity. The
backward regression method used in this study reduced the problem of multicollinearity.
Still, the individual regression coefficients should be interpreted with caution.
External validity is the extent to which the results of a study can be generalized
across populations, settings, and time. This study has low external validity. The predictor
variables that emerged as significant for the post-communist countries may not be as
significant for another set of countries or another region of the world. However, it is
important to note that these predictor variables would still be applicable for another group
of countries, but their relative effects on Internet use may vary.
The external validity is also low in terms of applying this framework to a different
point in time. It is expected that other variables may become more relevant to global
Internet adoption at different points in time. Structural variables such as income and
teledensity may be more significant during the early stages of Internet adoption whereas
the audience factors may become more critical at later stages. Socioeconomic factors tend
to be more important at earlier stages of Internet adoption. Also, income and
infrastructure as well as democratization level (the external factors) are likely to act as
stronger barriers to adoption than cultural predispositions.
The study captured Internet adoption at one point in time. Relationships between
variables may change at different stages of the adoption process. It is important to
underscore the dynamic nature of the Internet adoption process. Any research findings
become outdated quickly because ICT use is increasing dramatically in all regions of the
world (Rodriguez & Wilson, 2000; World Bank, 2001).
This study examined only country-level factors that affect Internet use. Other
factors may be influential at the individual or at the institutional level (Arquette, 2002;
European Commission, 2001; Norris, 2001). Norris (2001), for instance, proposed the
Internet Engagement Model, in which Internet adoption is to be studied at the macro
(country) level, the meso (institutional) level, and the micro (individual) level. Again, this
study focused only on the macro level and examined only country-level factors that affect
Even though the proposed model of Internet adoption was very comprehensive, it
did not include every single possible variable that may have an effect on Internet use.
Rather, the study explored the significance of a set of variables considered most
important for Internet adoption at the country level at this stage of the adoption process.
This study is mostly cross-sectional, which limits the applicability of the findings
for future stages of Internet diffusion. Internet diffusion is a process and not a static
phenomenon. However, this study captures the adoption process in the post-communist
countries at one point in time.
Reliability refers to the question if the same technique applied to the same data
yields the same results each time. Reliability does not seem to be an issue when using
secondary data as was the case in this study. There are no sampling issues involved
because the data gathering methods of the original sources are highly reliable. Since the
statistical methods were robust, we assert that the study has produced reliable results.
The interpretation of results in multiple regression always brings the question of
causality versus correlation. Clearly the correlation coefficients show a strong
relationship between the aforementioned independent variables and the dependent
variable, but is that relationship causal? Babbie (1995) lists three conditions that must be
present in order to claim a causal relationship: (1) time sequence; (2) correlation between
cause and effect; and (3) no other possible third variable that causes both the dependent
and independent variable.
Close examination of the variables in the current study shows that the first two
conditions are met. Let's take religion as an examine and test each of the three criteria.
Clearly, religion has come into existence before Internet use. Second, the correlation
between religion and Internet use is high, as determined by the dummy variable the
regression analysis. The third criterion is more difficult to test. Arguably, there is no
other variable that affects both religious predispositions and likelihood of Internet use.
Suggestions for Future Research
The Internet is one of the most fascinating new technologies of all time, and
research on its adoption around the world is growing. This dissertation explained cross-
country determinants of Internet adoption in the post-communist countries. Future
research should focus on adoption at the individual and organizational levels to see if
different determinants play a role. In addition, future research should examine if the
Internet is used differently by different individuals within the post-communist countries.
In other words, studies should identify individual demographic characteristics as well as
personality factors that affect adoption at the individual level in the post-communist
world. In this regard, more qualitative research of Internet adoption and use within the
post-communist countries is warranted.
Another venue for future research would be to isolate better proxy variables for the
broad set of factors suggested in this analysis. First, it would be desirable to incorporate
other cultural variables in addition to religion and test their effect on country-level
Internet adoption. Other measures of telecommunications policy such as level of
competition or Internet regulation should also be tested in future research. Future studies
of cross-country Internet diffusion should incorporate English language as an audience
factor. A better educational variable that should be tested as a determinant of Internet
adoption in the future is some specific measure of ICT skills in particular.
Future studies should apply the five-dimensional framework proposed here to a
world model of Internet adoption. It will be interesting to see if there are any regional
differences in the effects of the predictor variables when applied to all countries. Such
studies may help isolate a set of variables that are universally important in the Internet
This was mostly a cross-sectional study of Internet use in the post-communist
countries of Eastern Europe and the former Soviet Union. Future studies should also
examine the adoption process in the region longitudinally. A comparative study between
Western and Eastern European countries is another venue for future research. Such a
comparative study would possibly allow researchers to identify additional cultural
predispositions that influence Internet adoption. Finally, a comparative study between
Western and Eastern European countries longitudinally should also be conducted.
The importance of the Internet for national development is unquestionable. Today
the significance of communication technology is more crucial than ever. The main
message in a recent World Bank report reads:
Although traditional channels of communication will remain important, the new
information and communications technologies hold great potential for broadly
disseminating knowledge at low cost, and for reducing knowledge gaps both within
countries and between industrial and developing countries. (World Bank, 1999, 56)
This potential of the Internet makes it crucial to understand the driving forces of its
adoption at the country level.
The transition countries of Eastern Europe and the former Soviet Union have
undergone a unique period of transition since the end of the Cold War. This transition has
proved to be all but easy. A Freedom House report discusses the difficult transition to
building a civil society, democracy and market economy in the region as a whole
(Karatnycky et al., 1997). The United Nations also underscores the dramatic
transformation in the post-communist countries and notes the various costs of transition
from totalitarianism to capitalist democracy (UNDP, 1999). The transition has led to
social problems, economic downfall, inflation, and deteriorating educational system,
among other issues (UNDP, 1999). Transition costs have been so high that some have
suggested it is more appropriate to call it the great depression rather than a transition
When discussing the multidimensional nature of the transition process, it is
important to realize that the transition in the former Soviet bloc will not be completed
overnight (Fischer et al., 1998). On the contrary, "overcoming institutional legacies and
building new, effective institutions is a process that, by definition, will take a long time"
(Karatnycky et al., 1997, 21). Increased Internet access and use—even though hardly a
panacea— may contribute to making these societies reach their transition goals faster.
Higher Internet use may be used to help their development during the challenging
What this research has shown is that building a more democratic society leads to
increased Internet use. Democratization emerges as more important determinant of
Internet adoption than ever before suggested. However, the structural issues of income
and infrastructure also remain critical for Internet development in the region. Countries
with lower national income and poor telephone infrastructure have to overcome these
barriers in order to increase Internet use. But first, a necessary step to enlarging the
information superhighway in the post-communist countries is to open up their societies
and increase the political rights and freedoms of their citizens.
LIST OF REFERENCES
Agresti, A., & Finley, B. (1997). Statistical methods for the social sciences (3 rd ed.).
Upper Saddle River, NJ: Prentice Hall.
Ahmann, L. (1998). Internet access and political participation: Can the Internet play a
role in strengthening democracy in South Africa ? University of Florida:
Unpublished Master's Thesis.
Arnum, E., & Conti, S. (1998). Internet development worldwide: The new superhighway
follows the old wires, rails, and roads. Retrieved February 17, 2003, from http://
Arquette, Toby J. (2002). Social discourse, scientific method, and the digital divide:
Using the Information Intelligence Quotient (IIP) to generate a multi-layered
empirical analysis of digital division. Northwestern University: Unpublished Ph.D.
Dissertation. Retrieved February 17, 2003, from http://www.sla.purdue.edu/people/
Atkin, D., Jeffres, L., & Neuendorf, K. (1998). Understanding Internet adoption as
telecommunications behavior. Journal of Broadcasting & Electronic Media , 42(4),
Atkinson, R. D., & Court, R. H. (1998, November). The new economy index:
Understanding America's economic transition. Retrieved February 17. 2003, from
Babbie, E. (1995). The practice of social research (7 th ed.). Belmont, CA: Wadsworth.
Baily, M. N. (2001). U.S. economic performance and the challenge for Europe . Retrieved
February 17, 2003, from http://www.iie.com/papers/baily0601.htm
Barua, A., Pinnell, J., Shutter, J., & Whinston, A. B. (1999). Measuring the Internet
economy. Retrieved February 17, 2003, from http://cism.bus.utexas.edu/works/
Bazar, B., & Boalch, G. (1997). A preliminary model of Internet diffusion within
developing countries. Proceedings of the AUSWEB97 Conference, Southern Cross
University, Gold Coast, Australia. Retrieved May 7, 2000, from http://ausweb.scu
Bauer, J. M. (1994). The emergence of global networks in telecommunications:
Transcending national regulation and market constraints. Journal of Economic
Issues , 28(2), 391-402.
Beilock, R., & Dimitrova, D. V. (2003). An exploratory model of inter-country Internet
diffusion. Telecommunications Policy , 27(3-4), 237-252.
Berg-Schlosser, D., & Siegler, R. (1990). Political stability and development: A
comparative analysis of Kenya, Tanzania, and Uganda . London: L. Rienner
Berners-Lee, T. (1999). Weaving the Web: The original design and ultimate destiny of
the World Wide Web by its inventor . San Francisco: Harper Collins.
Bieler, D., & Stevenson, I. (1998, December). OVUM Report: Internet market forecasts:
Global Internet growth, 1998-2005. Retrieved February 17, 2003, from http://www
Bruce, R. R. (1999). Overcoming obstacles to liberalization of the telecom sector in
Estonia, Poland, the Czech Republic, Slovenia, and Hungary . Washington, D.C.:
The World Bank.
Campbell, R. W. (1995) Soviet and post-Soviet telecommunications: An industry under
reform . Boulder, CO: Westview Press.
Canning, A. (1997). Privatization and competition in Hungarian telecommunications. In
D. Ryan (ed.), Privatization and competition in telecommunications (pp. 103-126).
Westport, CN: Praeger.
Caselli, F., & Coleman II, W. J. (2000). The world technology frontier. Retrieved
February 17, 2003, from http://papers.nber.org/papers/w7904
Caselli, F., & Coleman II, W. J. (2001). Cross-country technology diffusion: The case of
computers. American Economic Review , 91(2), 328-335.
Castells, M. (1996). The rise of the network society . Cambridge, MA: Blackwell.
Castells, M. (2001). The Internet galaxy: Reflections on the Internet, business, and
society . Oxford: Oxford University Press.
Center for Democracy and Technology. (2000). Bridging the digital divide: Internet
access in Central and Eastern Europe . Retrieved February 17, 2003, from http://
Christensen, C. (1997). The innovator's dilemma: When new technologies cause great
firms to fail. Boston: Harvard Business School Press.
Christensen, C, Craig, T., & Hart, S. (2001). The great disruption. Foreign Affairs , 80(2),
Central Intelligence Agency. (2000). The world factbook . Retrieved February 17, 2003,
Clarke, G. R. G. (2001, July). Bridging the Digital Divide: How enterprise ownership and
foreign competition affect Internet access in Eastern Europe and Central Asia.
World Bank Working Paper 2629, Retrieved February 17, 2003, from
Colin Xu, L., Li, W., & Zhen-Wei Qiang, C. (2001). The political economy of
privatization and competition: Cross-country evidence from the
telecommunications sector. CEPR Discussion Paper 2825. Retrieved February 17,
2003, from http://ideas.repec.Org/p/cpr/ceprdp/2825.html
Comer, D. E. (1995). Internetworking with TCP/IP (3 rd ed.). Englewood Cliffs, NJ:
Cortez, M. V. (2000). Internet censorship around the world. Retrieved February 17, 2003,
Cyberatlas. (2002). Worldwide Internet population. Retrieved February 17, 2003, from
Daly, J. A. (2000, September). Will the Internet promote democracy? iMP Magazine ,
Retrieved February 17, 2003, from http://www.cisp.org/imp/september_2000/
Daly, J. A. (1999, May). Measuring the impacts of the Internet in the developing world.
iMP Magazine , Retrieved February 17, 2003, from http://www.cisp.org/imp/
Daly, J., & Miller, R. (1998). Corporations' use of the Internet in developing countries.
Washington, D.C.: The World Bank.
Dasgupta, S., Lall, S., & Wheeler, D. (2001, March 28). Policy reform, economic growth,
and the digital divide: An econometric analysis. World Bank Working Paper 2567,
Retrieved February 17, 2003, from http://econ.worldbank.org/view
.php?topic= 1 4&type=5&id= 1615
de Melo, M., & Gelb, A. (1996). A comparative analysis of twenty-eight transition
economies in Europe and Asia. Post-Soviet Geography and Economics , 37(5), 265-
DePrince, A. E., Jr., & Ford, W. F. (1999). A primer on internet economics: Macro and
micro impact of the Internet on the economy. Business Economics , 34(4), 42-50.
DiMaggio, P. (1997). Culture and cognition. Annual Review of Sociology , 23, 263-287.
Dimitrova, D. V. (2002). Internet uses and gratifications: An online survey of Bulgarians
at home and abroad. International Communication Bulletin , 37(1-2), 36-49.
Dinkova, D. (1998). Bulgaria's Internet: Nonprofit organizations are at the forefront.
Economic Reform Today , 3.
Domanski, H. (2000). On the verge of convergence: Social stratification in Eastern
Europe . New York: Central European University Press.
Drohan, M, & Freeman, A. (1997). Winning the language wars: The world speaks.
World Press Review , 6-8.
Dryden, J. (1998). Realising the potential of global electronic commerce. The OECD
Observer , 214. Retrieved February 17, 2003, from http://oecd.org/publications/
observer/2 14/ Article6_eng.htm
El-Nawawy, M. A. (2000). Profiling Internet users in Egypt: Understanding the primary
deterrent against their growth in number. Retrieved February 17, 2003, from http://
Elie, M. (1998). The Internet and global development. Retrieved February 17, 2003, from
Ellis, F. (1999). From glasnost to the internet: Russia's new infosphere . New York: St.
Estache, A., Manacorda, M., & Valletti, T. M. (2002, March 21). Telecommunication
reforms, access regulation, and Internet adoption in Latin America. World Bank
Working Paper 2802, Retrieved February 17, 2003, from http://econ.worldbank
.org/view.php?type=5&id= 1 3 1 62
European Bank for Reconstruction and Development. (1997). Transition report 1997:
Enterprise performance and growth . London: European Bank for Reconstruction
European Commission. (2000). Key data on education in Europe, 1999-2000 .
Luxembourg: Eurostat Press Office.
European Commission. (2001, March). The digital divide: A research perspective. Report
EUR 19913. Retrieved February 17, 2003, from http://www.fraggersxtreme.com/
lemon/eur 1 99 1 3en.pdf
Fischer, S., Sahay, R., & Vegh, C. A. (1998, April). How far is Eastern Europe from
Brussels? IMF Working Paper WP/98/53. Retrieved February 17, 2003, from http://
Fish, M. S. (1998). The determinants of economic reform in the post-communist world.
East European Politics & Societies , 12 (1), 31-78.
Forrester Research. (2000, August 15). Latin culture and climate explains low Internet
adoption in France, Italy, and Spain. Retrieved February 17, 2003, from http://www
Freedom House. (2000). Freedom House country ratings. Retrieved February 17, 2003,
Garrison, B. (2000). Online information use in newsrooms: A longitudinal diffusion
study. Paper presented to the Newspaper Division, Association for Education in
Journalism and Mass Communication (AEJMC) conference, Phoenix.
Global Reach. (2000, September). Global Internet statistics. Retrieved February 17, 2003,
Godwin, M. (1998). Cyber rights: Defending free speech in the digital age . New York:
Goode, S., & Stevens, K. (2000). An analysis of the business characteristics of adopters
and non-adopters of World Wide Web technology. Information Technology and
Management , 1, 129-154.
Gospic, N., Jankovic, M., & Odadzic, B. (2000, August). Yugoslav telecommunications
markets: Vision and Potential. IEEE Communications Magazine , 38(8), 112-116.
Gray, A., & McGuigan, J. (eds.). (1997). Studying culture: An introductory reader (2 nd
ed.). London: Arnold.
Gruber, H. (2001). Competition and innovation: The diffusion of mobile
telecommunications in Central and Eastern Europe. Information Economics and
Policy , 13, 19-34.
Guillen, M. & Suarez, S. (2001). Developing the Internet: Entrepreneurship and public
policy in Ireland, Singapore, Argentina and Spain. Telecommunications Policy , 25,
Gujarati, D. (1995). Basic econometrics (3 rd ed.). New York: McGraw-Hill.
Gulyas, A. (1998). In the slow lane on the information superhighway: Hungary and the
information revolution. Convergence: The Journal of Research into New Media
Technologies , 4. Retrieved February 17, 2003, from http://www.cios.org
Gunarante, S. A. (2001). Global triadization: A theoretical framework for global
communication research. Paper presented to the Communication Theory and
Methodology Division, Association for Education in Journalism and Mass
Communication (AEJMC) conference, Washington, D.C.
Guthrie, R.A., & Austin, L. D. (1996). Competitive implications of the Internet.
Information Systems Management , 13(3), 90-91.
GVU Web Surveys. (1998, October). Retrieved February 17, 2003, from
Hafner, K., & Lyon, M. (1996). Where wizards stay up late: The origins of the Internet .
New York: Simon & Schuster.
Hanson, J., & Narula, U. (1990). New communication technologies in developing
countries . Hillsdale, NJ: Lawrence Erlbaum Associates.
Hargittai, E. (1999). Weaving the Western web: Explaining differences in Internet
connectivity among OECD countries. Telecommunications Policy , 23(10/1 1), 701-
Held, D. (1995). Democracy and the global order: From the modern state to
cosmopolitan governance . Stanford, CA: Stanford University Press.
Hoelschner, G. (2000, August). Profile of the Czech communications market. IEEE
Communications Magazine , 38(8), 77-80.
Hofstede, G. (1980). Culture's consequences: International differences in work-related
values . Beverly Hills, CA: Sage.
Hofstede, G. (2001). Culture's consequences: Comparing values, behaviors, institutions,
and organizations across nations (2 nd ed.). Thousand Oaks, CA: Sage.
Horvath, J. (2002). Trouble in cyberspace. Telepolis. Retrieved February 17, 2003, from
http://www.telepolis.de/english/inhalt/te/ 1 2520/ 1 .html
Howard, P.E.N., L. Rainie, & Jones, S. (2001). Days and nights on the Internet: The
impact of a diffusing technology. American Behavioral Scientist , 45(3), 383-404.
International Monetary Fund. (2000a). World economic outlook: Asset prices and the
business cycle . Retrieved February 17, 2003, from http://www.imf.org
International Monetary Fund. (2000b). Dissemination standards bulletin board. Retrieved
February 17, 2003, from http://dsbb.imf.org/
International Telecommunication Union. (1999). Challenges to the network: Internet for
development . Geneva: ITU.
International Telecommunication Union. (2000). World telecommunication indicators .
Retrieved February 17, 2003, from http://www.itu.ch/ti/
International Telecommunication Union. (2001). Telecommunication regulation and
privatization data: Country profiles. Retrieved February 17, 2003, from
Jamison, M. (1995). A competitive framework for pricing interconnection in a global
telecommunications market. Denver Journal of International Law and Policy ,
Jasinski, P. (1997). Competition rules and regulations in telecommunications: The case of
Poland's intent to join the EU. In D. Ryan (ed.), Privatization and competition in
telecommunications (pp. 127-148). Westport, CN: Praeger.
Jisi, W., Yunus, M., Somavia, J., & Mesquita, R. L. (2001). Crossing the
digital divide: The Internet in China: A new fantasy? New Perspectives
Quarterly , 18(1), 22-24.
Jones, H. B. (1997, July). The Protestant ethic: Weber's model and the empirical
literature. Human Relations , 50(7), 757-798.
Jupiter Research Press Release. (2001a, January 1 1). Retrieved February 17, 2003, from
http://www.jmm.com/xp/jmm/press/2001/pr_01 1 lOl.xml
Jupiter Research Press Release. (2001b, January 18). Retrieved February 17, 2003, from
Karatnycky, A., Motyl, A., & Shor, B. (eds.). (1997). Nations in transit: 1997: Civil
society, democracy and markets in East Central Europe and the Newly Independent
States . New Brunswick: Transaction Publishers.
Katchanovski, I. (2000). Divergence in growth in post-communist countries. Journal of
Public Policy , 20(1), 55-81.
Katz, J. E., Rice, R. E., & Aspden, P. (2001). The Internet, 1995-2000: Access, civic
involvement, and social interaction. American Behavioral Scientist , 45(3), 405-419.
Kennedy, P. (1998). A guide to econometrics (4 th ed.). Cambridge, MA: The MIT Press.
Kiiski, S., & Pohjola, M. (2001, June). Cross-country diffusion of the Internet.
UNU/WIDER Discussion Paper 2001/1 1. Retrieved February 17, 2003, from
www.wider.unu.edu/publications/dps/DP2001-l 1 .pdf
Kleinbaum, D. G., Kupper, L. L., & Muller, K. E. (1998). Applied regression analysis
and other multi variable methods . Boston, MA: PWS-KENT Publishing Co.
Kouznetsov, A., & Bourtsev, D. (1996). Prospects for the development of the Internet in
Russia. INET Conference Proceedings . Retrieved February 17, 2003, from http://
Kuentzel, D., Sloutski, L., & Sokolov, N. (2000, December). Evolution of
telecommunication in Eastern Europe. IEEE Communications Magazine , 38(8),
Lamberton, D. (1997). The new research frontiers of communications policy .
Amsterdam, The Netherlands: Elsevier.
Lari, E. F. (2000). International institutions in Eastern Europe: Into the financial breach.
Harvard International Review , 13(1), 18.
Leiner, B. M, Vinton G. Cerf, V. C, Clark, D. D., Kahn, R. E., Kleinrock, L., Lynch, D.
C, Postel, J., Roberts, L. G., & Wolff, S. (1997). A brief history of the Internet.
Retrieved February 17, 2003, from http://www.isoc.org/internet/history/brief.html
Lin, C. (1998). Exploring personal computer adoption dynamics. Journal of Broadcasting
& Electronic Media , 42(1), 95-112.
Lin, C. (1999). Online-service adoption likelihood. Journal of Advertising Research ,
Lin, N. (1993). Diffusion of information technology: A case study of computer network
and the role of government, industry, and academia in developing the
Internet/NREN . University of Texas Austin: Unpublished Ph.D. Dissertation.
Lindstrom, P. (1997). The Internet: Nielsen's longitudinal research on behavioral changes
in use of this counterintuitive medium. Journal of Media Economics , 10(2), 35-40.
Maddala, G. (1984). Limited-dependent and qualitative variables in econometrics .
Cambridge: Cambridge University Press.
Maddock, R. (1997). Telecommunications and economic development. In D. Lamberton
(ed.). The new research frontiers of communications policy (pp. 159-175).
Amsterdam, The Netherlands: Elsevier.
Madon, S. (2000). The Internet and socio-economic development. Information
Technology and People , 13(2), 85-101.
Magyar, B., & Karvalics, L. Z. (2001). "Information society" in Eastern Europe?
Chances, possibilities, tasks and programs. East European Quarterly , XXXIV, 4.
Maherzi, A. (1997). World communication report: The media and the challenge of the
new technologies . Paris: UNESCO Publishing.
Mahler, A., & Rogers, E. (1999). The diffusion of interactive communication innovations
and the critical mass: The adoption of telecommunications services by German
banks. Telecommunications Policy , 23, 719-740.
Maitland, C. (1998). Global diffusion of interactive networks: The impact of culture.
Electronic Journal of Communication , 8. Retrieved February 17, 2003, from
Malecki, E. J. (1997). Technology and economic development: The dynamics of local,
regional and national competitiveness (2 nd ed.). Harlow, England: Longman.
Malecki, E. J. (2000). Knowledge and regional competitiveness. Erdkunde , 54(4), 334-
Malecki, E. (2001). The Internet age: Not the end of geography. In Felsenstein, D., &
Taylor, M.J. (eds.) Promoting local growth: Process, practice and policy .
McChesney, R. (1999). Rich media, poor democracy: Communication politics in dubious
times . Urbana: University of Illinois Press.
McElhinney, S. (2001). Telecommunications liberalisation and the quest for universal
service in Australia. Telecommunications Policy , 25, 233-248.
McKnight, L. W., & Bailey, J. (1997). Internet economics . Cambridge, MA: MIT Press.
McLuhan, M., & Powers, B. R. (1989). The global village: Transformations in world life
and media in the 21 st century . New York: Oxford University Press.
Media Metrix Global Services. (2000). Retrieved February 17, 2003, from http://www
Mendoza, M., & Alvarez de Toledo. (1997). Demographics and behavior of the Chilean
Internet population. Journal of Computer-Mediated Communication , 3(1).
Retrieved February 17, 2003, from http://www.ascusc.org/jcmc/vol3/issuel/
Michalis, M., & Takla, L. (1997). Telecommunications in the Czech Republic: The
privatization of SPT Telecom. In D. Ryan (ed.), Privatization and competition in
telecommunications (pp. 89-102). Westport, CN: Praeger.
Minges, M. (2000). Counting the Net: Internet access indicators. Retrieved February 17,
2003, from http://www.isoc.org/inet2000/cdproceedings/8e/8e_l.htm
Minges, M. (2001). Internet around the world. Retrieved February 17, 2003, from
Mitchell, W. J. (1995). City of bits: Space, place, and the infobahn . Cambridge: MIT
Nelson, R. R. (1993). National innovation systems . New York: Oxford University Press.
Network Wizards Internet Domain Survey. (2000). Retrieved February 17, 2003, from
Newhagen, J., & Rafaeli, S. (1996). Why communication researchers should study the
Internet: A dialogue. Journal of Communication , 46, 4-13.
Nielsen NetRatings. (2000). Retrieved February 17, 2003, from http://www.nielsen-
Nielsen NetRatings. (2002). Retrieved February 17, 2003, from http://www.nielsen-
Norris, P. (2000). The worldwide digital divide: Information poverty, the Internet and
development. Paper presented at the annual meeting of the Political Studies
Association, London School of Economics and Political Science, UK. Retrieved
February 17, 2003, from http://www.ksg.harvard.edu/iip/governance/psa2000dig
Norris, P. (2001). Digital divide: Civic engagement, information poverty and the Internet
in democratic societies . New York: Cambridge University Press.
National Telecommunications and Information Administration. (1995, July). Falling
through the Net: A survey of the "have nots" in rural and urban America. Retrieved
February 17, 2003, from http://www.ntia.doc.gov/ntiahome/fallingthru.html
National Telecommunications and Information Administration. (1998, July). Falling
through the Net II: New data on the digital divide. Retrieved February 17, 2003,
National Telecommunications and Information Administration. (1999, July). Falling
through the Net: Defining the digital divide. Retrieved February 17, 2003, from
Nua's Internet user surveys. (2002, September). Retrieved February 17, 2003, from http://
Oaca, N. (2000, August). Mobile telephony: The main driver of Romanian
telecommunications! IEEE Communications Magazine , 38(8), 98-104.
Organization for Economic Cooperation and Development. (1998a, October). Internet
infrastructure indicators. Retrieved February 17, 2003, from http://www.oecd.org/
Organization for Economic Cooperation and Development. (1998b). Internet traffic
exchange: Developments and policy . Retrieved February 17, 2003, from http://
Paltridge, S. (2000). Local access pricing and the international digital divide. Retrieved
February 17, 2003, from http://www.isoc.org/oti/articles/1000/paltridge.html
Papir, Z., & Oleszak, P. (2000, August). The communications market in Poland. IEEE
Communications Magazine , 38(8), 91-95.
Perrit, H. H. (1999, February). The Internet as a threat to sovereignty? Thoughts on the
Internet's role in strengthening national and global governance. Retrieved February
17, 2003, from http://www.law.indiana.edu/glsj/vol5/no2/4perrit.html
Petrazzini, B., & Guerrero, A. (2000). Promoting Internet development: The case of
Argentina. Telecommunications Policy , 24(2), 89-1 12.
Pew Center for the People and the Press. (1995). Technology in the American Household .
Washington, D.C.: Pew Center for the People and the Press.
Pitkow, J. (1996). Emerging trends in the WWW user population. Communications of the
ACM , 39(6), 106-108.
Poster, M. (1995). CyberDemocracy: Internet and the public sphere. Retrieved February
17, 2003, from http://www.hnet.uci.edu/mposter/writings/democ.html
Poster, M. (2001). What's the matter with the Internet . Minneapolis, MN: University of
Prescott, M. B., & Van Slyke, S. (1997). Understanding the Internet as an innovation.
Industrial Management and Data Systems , 97(3), 119-124.
Press, L. (1997, November). Tracking the global diffusion of the Internet.
Communications of the ACM , 40(1 1), 11-17.
Press, L., Burkhart, G., Foster, W., Goodman, S., Wolcott, P., & Woodard, J. (1998). An
Internet diffusion framework. Communications of the ACM , 41(10), 21-26.
Pritchett, L., & Kaufmann, D. (1998, March). Civil liberties, democracy, and the
performance of government projects. Finance & Development . Retrieved February
17, 2003, from http://www.worldbank.org/fandd/english/pdfs/0398/0140398.pdf
Rey, L. (1998). Multiculturality and communication technologies in Switzerland.
Electronic Journal of Communication , 8. Retrieved February 17, 2003, from
Research on Internet in Slovenia. (2000). Retrieved February 17, 2003, from http://www
Rodriguez, F., & Wilson, E. J. (2000, May). Are poor countries losing the information
revolution? Retrieved February 17, 2003, from http://www.infodev.org/library/
Rogers, E. (1995). Diffusion of innovations (4 th e<±). New York: Free Press.
Rogerson, K., & Thomas, D. G. (1998). Internet regulation process model: The effect of
societies, communities, and governments. Journal of Political Communication , 15,
Roller, L., & Waverman, L. (2001). Telecommunications infrastructure and economic
development: A simultaneous approach. American Economic Review , 91(4), 909-
Romer, P. (1999). What makes technology grow? The Wilson Quarterly , 23(2), 1 1-13.
Rood, M. (1999). A word about Internet statistics. Telecommunications Policy ,
Rose, R. (2002). Digital divide or digital diffusion? Transition , 13(4-5), 33-35.
Ryan, D. (1997). Privatization and competition in telecommunications . Westport, CN:
Ryan, B., & Gross, N. C. (1943). The diffusion of hybrid seed corn in two Iowa
communities. Rural Sociology , 8, 15-24.
Sadowsky, G. (1993). Network connectivity for developing countries. Communications
of the ACM , 36(8), 42-47.
Sale, K. (1999). The triumph of technology. The Ecologist , 29(3), 187-188.
Sallai, G. (2000, August). Reform and development of Hungarian telecommunications.
IEEE Communications Magazine , 38(8), 82-87.
Sen, A. (1999a). Democracy as a universal value. Journal of Democracy , 10(3), 3-17.
Sen, A. (1999b). Development as freedom . New York: Knopf.
Severin, W. J., & Tankard, J. W., Jr. (1997). Communication theories: Origins, methods,
and uses in the mass media (4 th ed.). New York: Longman.
Shapiro, C, & Varian, H. R. (1999). Information rules: A strategic guide to the network
economy . Boston, MA: Harvard Business School Press.
Singh, J. P. (1999). Leapfrogging development? The political economy of
telecommunications restructuring . Albany, NY: State University of New York
Sokolov, N., & Goldenstein, B. (2000, August). Telecommunications in Russia. IEEE
Communications Magazine , 38(8), 106- 111.
Solomon, N. (1998). Internet shopping network: The mailing of cyberspace. Retrieved
February 17, 2003, from http://www.fair.org/media-beaty981029.html
Sondergaard, M. (1994). Hofstede's consequences: A study of reviews, citations and
replications. Organization Studies , 15(3), 447- 456.
Standage, T. (1998). The Victorian internet: The remarkable story of the telegraph and
the 19 th century's on-line pioneers . New York: Walker.
Stevens, J. (1992). Applied multivariate statistics for the social sciences (2 n ed.).
Hillsdale, NJ: Lawrence Erlbaum.
Stewart, C. M., Shields, S. F., & Sen, N. (1998). Diversity in on-line discussions: A
study of cultural and gender differences in listservs. Electronic Journal of
Communication , 8. Retrieved February 17, 2003, from http://www.cios.org/getfile
Tanner, E. (1999). Links to the world. Gazette: International Journal for Communication
Studies , 61(1), 39-57.
Tarde, G. (1903). The laws of imitation . New York: H. Holt and Company.
Tayeb, M. (1994). Organizations and national culture: Methodology considered.
Organization Studies , 15(3), 429-446.
Tele-haves and have-nots: Developing countries make use of technological innovations.
(1996, May 18). The Economist , 19-20.
The cyber challenge. (2000, July 3). Transitions Online . Retrieved February 17, 2003,
The Internet's new borders. (2001, August 11). The Economist , 9-10.
The new economy. (2000, September 23). The Economist , 5-40.
Tobin, J. (1958). Estimation of relationships for limited dependent variables.
Econometrica , 26, 24-36.
United Nations Development Programme. (1999). Transition 1999: Human development
report for Central and Eastern Europe and the CIS . New York: UNDP.
United States Agency for International Development. (2000). Bulgaria assessment:
Internet environment for economic development. Retrieved February 17, 2003,
United States Census Bureau. (1997, October). Computer use and ownership. Retrieved
February 17, 2003, from http://www.census.gov/population/www/socdemo/
United States Commerce Department. (1998, April). The emergin g digital economy.
Retrieved February 17, 2003, from http://www.ecommerce.gov/emerging.htm
United States Internet Council. (2000, September). State of the Internet 2000. Retrieved
February 17, 2003, from http://www.usic.org/papers/stateoftheinternet2000/
Wallraff, B. (2000, November). What global language? The Atlantic Monthly , 52-66.
Wallsten, S. (1999). An empirical analysis of competition, privatization, and regulation in
telecommunications markets in Africa and Latin America. World Bank Working
Paper 2817, Retrieved February 17, 2003, from http://ideas.repec.Org/p/wop/
Wallsten, S. (2002, March 25). Does sequencing matter? Regulation and privatization in
telecommunications reforms. World Bank Working Paper 2817, Retrieved
February 17, 2003, from http://econ.worldbank.org/view.php?type=5&id= 13266
Weir, T. (1999). Innovators or news hounds? Newspaper Research Journal , 20(4), 62-81.
Wheatley, J. (1999). World telecommunications economics . London: The Institution of
Wilson, B., Ryder, M., McCahan, J., & Sherry, L. (1996). Cultural assimilation of the
Internet: A case study. Retrieved February 17, 2003, from http://carbon.cudenver
Winner, L. (1997) Technology today: Utopia or dystopia? Social Research , 64(3), 989-
Winston, B. (1998). Media technology and society: A history: From the telegraph to the
Internet . London: Routledge.
Wolcott, P., Press, L., McHenry, W., Goodman, S., & Foster, W. (2001). A framework
for assessing the global diffusion of the Internet. Journal of the Association for
Information Systems , 2(6). Retrieved February 17, 2003, from http://www.isns
World Bank. (1998). World development indicators . Washington, D.C.: The World Bank.
World Bank. (2000). World development indicators . Washington, D.C.: The World Bank.
World Bank. (2001). World development indicators . Washington, D.C.: The World Bank.
World Intellectual Property Organization. (2001). Primer on electronic commerce and
intellectual property issues. Retrieved February 17, 2003, from http://ecommerce
. wipo.int/primer/section 1 .html
Xavier, P. (2000). Market liberalisation and regulation in Hungary's telecommunications
sector. Telecommunications Policy , 24(10/11), 807-841.
Yakovlev, Y. (1989). Flagship of Glasnost. In Cohen S., & vanden Heuvel, K. (eds.)
Voices of Glasnost: Interviews with Gorbachev's Reformers (pp. 197-212). New
Zook, M. (2000). Internet metrics: Using host and domain counts to map the internet.
Telecommunications Policy , 24(6/7), 613-620.
Zook, M. (2002). Zooknic Internet intelligence: Internet users worldwide. Retrieved
February 17, 2003, from http://www.zooknic.com/Users/index.htm
Dimitrova received a Master of Arts in journalism and communication from the
University of Oregon in 1999. She holds a Bachelor of Arts degree in journalism/mass
communication and political science/international relations from the American University
in Bulgaria (AUBG). Dimitrova' s professional background is in radio and television
Daniela Dimitrova conducts research in the area of new media and the Internet. Her
research interests include Internet use at both the micro and macro level. She has
examined the content and design of online media in previous studies, focusing on the use
of hyperlinks, multimedia and interactivity. In addition, Dimitrova' s research focuses on
political communication and online media management.
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
Sylvjt Chan-Olmsted, Chair
Associate Professor of Journalism and
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy .^-
Professor of Journalism and
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
f A.: ^ 'Tacm^-
Professor of Journalism and
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
Associate Professor of Journalism and
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
Professor of Food and Resource Economics
This dissertation was submitted to the Graduate Faculty of the College of
Journalism and Communications and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
May 2003 A>
san, College4)f Journalism and
Dean, Graduate School
UNIVERSITY OF FLORIDA
3 1262 08557 1932